
# import all packages and set plots to be embedded inline
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

%matplotlib inline
np.random.seed(123) # for reproducing same random results


#reading 2008.csv then copying to a dataframe
flights_2008 = pd.read_csv('2008.csv')
df = flights_2008.copy()
print(df.shape)
print(df.info())
df.head()

(7009728, 29)
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7009728 entries, 0 to 7009727
Data columns (total 29 columns):
 #   Column             Dtype  
---  ------             -----  
 0   Year               int64  
 1   Month              int64  
 2   DayofMonth         int64  
 3   DayOfWeek          int64  
 4   DepTime            float64
 5   CRSDepTime         int64  
 6   ArrTime            float64
 7   CRSArrTime         int64  
 8   UniqueCarrier      object 
 9   FlightNum          int64  
 10  TailNum            object 
 11  ActualElapsedTime  float64
 12  CRSElapsedTime     float64
 13  AirTime            float64
 14  ArrDelay           float64
 15  DepDelay           float64
 16  Origin             object 
 17  Dest               object 
 18  Distance           int64  
 19  TaxiIn             float64
 20  TaxiOut            float64
 21  Cancelled          int64  
 22  CancellationCode   object 
 23  Diverted           int64  
 24  CarrierDelay       float64
 25  WeatherDelay       float64
 26  NASDelay           float64
 27  SecurityDelay      float64
 28  LateAircraftDelay  float64
dtypes: float64(14), int64(10), object(5)
memory usage: 1.5+ GB
None


# finding duplicated data points
df[df.duplicated()]


# reduced dataset with 19 columns with carriers names after left joining carriers.csv via SQLite 

flights_sql = pd.read_csv('2008_flights.csv')
flights = flights_sql.copy()

# removing duplicated datapoints found within jupyter notebook before the SQLite work 

flights.drop(index=[938221,938223,938225,938227],inplace=True)
flights.reset_index(drop=True,inplace=True)
print(flights.shape)
print(flights.info())
flights.head()

(7009724, 19)
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7009724 entries, 0 to 7009723
Data columns (total 19 columns):
 #   Column            Dtype  
---  ------            -----  
 0   FlightNum         int64  
 1   TailNum           object 
 2   Month             int64  
 3   DayofMonth        int64  
 4   DayOfWeek         int64  
 5   Carrier           object 
 6   ArrDelay          float64
 7   Cancelled         int64  
 8   CancellationCode  object 
 9   Diverted          int64  
 10  CarrierDelay      float64
 11  WeatherDelay      float64
 12  NASDelay          float64
 13  SecurityDelay     float64
 14  Origin            object 
 15  Dest              object 
 16  Distance          int64  
 17  TaxiIn            float64
 18  TaxiOut           float64
dtypes: float64(7), int64(7), object(5)
memory usage: 1016.1+ MB
None


# extracting the columns thought as helpful for this investigating

investigate_cols = ['Month', 'Carrier', 'ArrDelay',
                    'Cancelled','CancellationCode',
                    'CarrierDelay','WeatherDelay','NASDelay','SecurityDelay']

df_inv = flights[investigate_cols].copy()
print(df_inv.shape)
print(df_inv.info())
df_inv.head()

(7009724, 9)
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7009724 entries, 0 to 7009723
Data columns (total 9 columns):
 #   Column            Dtype  
---  ------            -----  
 0   Month             int64  
 1   Carrier           object 
 2   ArrDelay          float64
 3   Cancelled         int64  
 4   CancellationCode  object 
 5   CarrierDelay      float64
 6   WeatherDelay      float64
 7   NASDelay          float64
 8   SecurityDelay     float64
dtypes: float64(5), int64(2), object(2)
memory usage: 481.3+ MB
None


#  replacing Months numbers with names                   ---- months
months = { 10:'Oct', 11:'Nov', 12:'Dec',1:'Jan', 2:'Feb', 3:'Mar',
            4:'Apr', 5:'May', 6:'Jun', 7:'Jul', 8:'Aug', 9:'Sep'}
df_inv['Month'] = df_inv['Month'].replace(list(months.keys()),
                                          list(months.values()))

# changing 'Month' type column to ordered category       ----- converting 'Month' to category
months_order = ['Jan','Feb','Mar','Apr','May','Jun',
                'Jul','Aug','Sep','Oct','Nov','Dec']
months_cat = pd.api.types.CategoricalDtype(ordered=True, categories=months_order)
df_inv['Month'] = df_inv['Month'].astype(months_cat)


# preparing for plots of Carrier for the whole data points df_inv (7,009,724 rows)
print('(df_inv carrier counts)','\n')
print(df_inv['Carrier'].value_counts().sort_index())

(df_inv carrier counts) 

AirTran Airways Corporation                                                             261684
Alaska Airlines Inc.                                                                    151102
Aloha Airlines Inc.                                                                       7800
American Airlines Inc.                                                                  604885
American Eagle Airlines Inc.                                                            490693
Atlantic Southeast Airlines                                                             280575
Comair Inc.                                                                             197607
Continental Air Lines Inc.                                                              298455
Delta Air Lines Inc.                                                                    451931
Expressjet Airlines Inc.                                                                374510
Frontier Airlines Inc.                                                                   95758
Hawaiian Airlines Inc.                                                                   61826
JetBlue Airways                                                                         196091
Mesa Airlines Inc.                                                                      254930
Northwest Airlines Inc.                                                                 347652
Pinnacle Airlines Inc.                                                                  262208
Skywest Airlines Inc.                                                                   567159
Southwest Airlines Co.                                                                 1201754
US Airways Inc. (Merged with America West 9/05. Reporting for both starting 10/07.)     453589
United Air Lines Inc.                                                                   449515
Name: Carrier, dtype: int64


# shortening long carrier name 'US Airways Inc. (Merged with America West 9/05. Reporting for both starting 10/07.)'
# within the whole data points dataframe df_inv  (7,009,728 rows)

long_str = 'US Airways Inc. (Merged with America West 9/05. Reporting for both starting 10/07.)'
short_str = 'US Airways Inc.'

df_inv['Carrier'] = df_inv['Carrier'].replace(long_str,short_str)


# exporting df_inv to 'csv' file
#df_inv.to_csv('2008_investigate_cols.csv',index = False)


# descriptive statistics for numeric variables
df_inv.describe()


# data points of cancelled flights           ************* cancelled flights 'airline_cancelled' (137,434 rows)
cancel_cols = ['Month','Carrier','CancellationCode']
airline_cancelled = df_inv.loc[df_inv['Cancelled']==1,cancel_cols].copy()
airline_cancelled.reset_index(drop=True,inplace=True)
print(airline_cancelled.shape)
print(airline_cancelled.info())
airline_cancelled.head()

(137434, 3)
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 137434 entries, 0 to 137433
Data columns (total 3 columns):
 #   Column            Non-Null Count   Dtype   
---  ------            --------------   -----   
 0   Month             137434 non-null  category
 1   Carrier           137434 non-null  object  
 2   CancellationCode  137434 non-null  object  
dtypes: category(1), object(2)
memory usage: 2.2+ MB
None


# descriptive statistics for variables of cancelled flights
airline_cancelled.describe()


# on-time flights, or data points with arrival delays less than 15 minutes and not null ** on-time (5,330,292 rows)
on_time = df_inv[(df_inv['ArrDelay'] < 15) & (~df_inv['ArrDelay'].isna())].copy()
on_time.reset_index(drop=True,inplace=True)
print(on_time.shape)
print(on_time.info())
on_time.head()

(5330292, 9)
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5330292 entries, 0 to 5330291
Data columns (total 9 columns):
 #   Column            Dtype   
---  ------            -----   
 0   Month             category
 1   Carrier           object  
 2   ArrDelay          float64 
 3   Cancelled         int64   
 4   CancellationCode  object  
 5   CarrierDelay      float64 
 6   WeatherDelay      float64 
 7   NASDelay          float64 
 8   SecurityDelay     float64 
dtypes: category(1), float64(5), int64(1), object(2)
memory usage: 330.4+ MB
None


# descriptive statistics for numeric variables with ArrDelay less than 15 minutes (on-time)
on_time.describe()


# extracting data points with arrival delays that equal 15 minutes or more
df_inv_15 = df_inv[df_inv['ArrDelay']>=15].copy()
df_inv_15.drop(columns=['Cancelled','CancellationCode'],inplace=True)

df_inv_15.reset_index(drop=True,inplace=True)
print(df_inv_15.shape)
print(df_inv_15.info())
df_inv_15.head()

(1524733, 7)
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1524733 entries, 0 to 1524732
Data columns (total 7 columns):
 #   Column         Non-Null Count    Dtype   
---  ------         --------------    -----   
 0   Month          1524733 non-null  category
 1   Carrier        1524733 non-null  object  
 2   ArrDelay       1524733 non-null  float64 
 3   CarrierDelay   1524733 non-null  float64 
 4   WeatherDelay   1524733 non-null  float64 
 5   NASDelay       1524733 non-null  float64 
 6   SecurityDelay  1524733 non-null  float64 
dtypes: category(1), float64(5), object(1)
memory usage: 71.3+ MB
None


df_inv_15.Month.unique()

['Jan', 'Feb', 'Mar', 'Apr', 'May', ..., 'Aug', 'Sep', 'Oct', 'Nov', 'Dec']
Length: 12
Categories (12, object): ['Jan' < 'Feb' < 'Mar' < 'Apr' ... 'Sep' < 'Oct' < 'Nov' < 'Dec']


# descriptive statistics for numeric variables with ArrDelay equals 15 minutes or more
df_inv_15.describe()


# sampling 500 flights so that plots are clearer and
# they render faster

samples = np.random.choice(df_inv_15.shape[0], 500, replace = False)
df_inv_15_samp = df_inv_15.loc[samples,:]


# univariate plots of CancellationCode or flight cancellation causes codes for the whole data points df_inv (7,009,728 rows)
# A = carrier, B = weather, C = NAS, D = security

plt.figure(figsize=(20,6))
default_color = sns.color_palette()[0]

ax = sns.countplot(data = df_inv, x = 'CancellationCode',
                  order = df_inv['CancellationCode'].value_counts().sort_index().index,
                  color = default_color,palette='muted')
ax.set_box_aspect(3.5/len(ax.patches)) # change 5 (numerator)  to modify the y/x axis ratio

sns.despine()

plt.title("Flights-cancellation causes \nfor year 2008",
          fontsize=18,
          color='#BF1B00',
          weight='bold')

plt.xlabel('Causes',fontsize=16.5,color='blue')
plt.ylabel('Freq.\n ',fontsize=16.5,color='blue')

plt.yticks([0,10000,20000,30000,40000,50000],
           [0]+[str(n)+'e4' for n in range(1,6)],#'1e4','2e4','3e4','4e4','5e4'],
           fontsize=12.5,weight='bold')

# Recalculating the causes_counts
causes_counts = df_inv['CancellationCode'].value_counts().sort_index()
# get the current tick locations and labels
locs, labels = plt.xticks(fontsize=12.5,weight='bold') 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):
    # get the text property for the label to get the correct count
    count = causes_counts[label.get_text()]
    # print the annotation just above the top of the bar
    plt.text(loc, count+250, s=f"{count:,}", ha = 'center', color='brown',weight='bold');
    
#ax.set_xticklabels(causes);
print('\n','Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security')

 Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security


# # univariate plots of CancellationCode or flight cancellation-causes codes within Months
# for the whole data points df_inv (7,009,728 rows)
# A = carrier, B = weather, C = NAS, D = security

sns.set(style='whitegrid')
g = sns.FacetGrid(data=df_inv,col='Month'
                  ,col_wrap=4
                 )#,xlim=(0,300))
g.fig.suptitle('Flights-cancellation causes frequencies per Months')

order = df_inv['CancellationCode'].value_counts().sort_index().index

g.map(sns.countplot,'CancellationCode', order = order,
                    hue=df_inv['CancellationCode'],hue_order= order,
                    color = default_color,palette='muted')
g.set_titles('{col_name}')
g.add_legend(labels=['Carrier', 'Weather', 'National Air System', 'Security'])
g.set_ylabels('Freq.');#,color='blue',fontsize=15);

print('\n','Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security')

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:132: UserWarning: You have mixed positional and keyword arguments, some input may be discarded.
  figlegend = self.fig.legend(handles, labels, **kwargs)

 Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security


# # univariate plots of CancellationCode or flight cancellation-causes codes within Carriers
# for the whole data points df_inv (7,009,728 rows)
# A = carrier, B = weather, C = NAS, D = security

sns.set(style='whitegrid')
g = sns.FacetGrid(data=df_inv,col='Carrier'
                  ,col_wrap=4
                 )#,xlim=(0,300))
g.fig.suptitle('Flights-cancellation causes frequencies per Carriers')

order = df_inv['CancellationCode'].value_counts().sort_index().index

g.map(sns.countplot,'CancellationCode', order = order,
                    hue=df_inv['CancellationCode'],hue_order= order,
                    color = default_color,palette='muted')
g.set_titles('{col_name}')
g.add_legend(labels=['Carrier', 'Weather', 'National Air System', 'Security'])
g.set_ylabels('Freq.');#,color='blue',fontsize=15);

print('\n','Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security')

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:132: UserWarning: You have mixed positional and keyword arguments, some input may be discarded.
  figlegend = self.fig.legend(handles, labels, **kwargs)

 Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security


# # univariate plots of CancellationCode or flight cancellation-causes codes within Months per Carriers 
# for the whole data points df_inv (7,009,724 rows)
# A = carrier, B = weather, C = NAS, D = security

sns.set(style='whitegrid')
g = sns.FacetGrid(data=df_inv,col='Month',row='Carrier',
                  #col_wrap=6,
                  margin_titles=True)#,xlim=(0,300))
g.fig.suptitle('Flights-cancellation causes frequencies per Months & Carriers',fontsize=24,x=0.385,y=1.025)

order = df_inv['CancellationCode'].value_counts().sort_index().index

g.map(sns.countplot,'CancellationCode', order = order,
                    hue=df_inv['CancellationCode'],hue_order= order,
                    color = default_color,palette='muted')

# # we can use 'catplot' instead of FacetGrid and g.map() like below

# g = sns.catplot(data=df_inv,kind='count',
#                 x='CancellationCode', order=order,
#                 col='Month',#col_wrap=6,
#                 row='Carrier',margin_titles=True,
#                 hue='CancellationCode',hue_order=order,palette = 'muted',
#                 size = 4)

g.set_titles(col_template='{col_name}',row_template='{row_name}')

g.set_ylabels('Freq.')
g.set(yticks = np.arange(0,40,5)*100)
yticklabels = [str(n/1000) + 'k' for n in g.axes[0,0].get_yticks()]
g.set_yticklabels(labels = [0] +yticklabels[1:],fontsize=15.5)

g.add_legend(labels=['Carrier', 'Weather', 'National Air System', 'Security'],
             title='Delay causes',#title_fontsize =25,
             loc='upper center',
             ncol=4,
             prop={'size':20},
             bbox_to_anchor=(0.385,1.013),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=22);


print('\n','Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security')

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:132: UserWarning: You have mixed positional and keyword arguments, some input may be discarded.
  figlegend = self.fig.legend(handles, labels, **kwargs)

 Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security


# univariate plots of ArrDelay arrival delay for the whole data points df_inv  (7,009,728 rows)
plt.figure(figsize=(10,8))

# histogram
plt.subplot(3,1,1)
sns.distplot(df_inv['ArrDelay'],kde=False,hist_kws={'histtype':'bar','alpha':0.1875})

# boxplot
plt.subplot(3,1,2)
sns.boxplot(data=df_inv , y='ArrDelay',color=sns.color_palette()[0])

# violinplot
plt.subplot(3,1,3)
sns.violinplot(data=df_inv , y='ArrDelay',color=sns.color_palette()[0]);

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\distributions.py:2557: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)


# finding arrival-delays outliers
delays = [2500,2000,1500,1000,500,400,350,250]

print(f"\nFrom {len(df_inv_15):,} flights of 15 minutes arrival delay or more:\n*****")
for delay in delays[::-1]:
    print(f"'{df_inv_15[df_inv_15.ArrDelay > delay].shape[0]:,}' flights",
          'exceeded "{}" min. arrival delay'.format(delay))
    

#     print('Flights that exceeded "{}" minutes arrival delay are'.format(delay),
#           f"'{df_inv_15[df_inv_15.ArrDelay > delay].shape[0]:,}' from",
#           f"{len(df_inv_15):,} flights of 15 minutes arrival delay or more.")

From 1,524,733 flights of 15 minutes arrival delay or more:
*****
'20,726' flights exceeded "250" min. arrival delay
'5,824' flights exceeded "350" min. arrival delay
'3,439' flights exceeded "400" min. arrival delay
'1,562' flights exceeded "500" min. arrival delay
'189' flights exceeded "1000" min. arrival delay
'9' flights exceeded "1500" min. arrival delay
'2' flights exceeded "2000" min. arrival delay
'0' flights exceeded "2500" min. arrival delay


# univariate plots of ArrDelay arrival delay for the whole data points df_inv_15 (1,524,733 rows)
# with limiting visualization up to 500 minutes

plt.figure(figsize=(15,15))

# histogram with kde distribution line
plt.subplot(3,1,1)
ax = sns.histplot(df_inv_15['ArrDelay'],bins=155,kde=True)#,hist_kws={'histtype':'bar','alpha':0.1875})
plt.xticks([0, 15, 30,50,78,100,115,130,150,200,215,230,250,300,315,330,350,400,500])
plt.xlim([0,500])
plt.yticks([0,100000,200000,300000,400000,500000,600000],
           [0,'1e5','2e5','3e5','4e5','5e5','6e5'])
#ax.set_xticklabels([],minor=True)

# boxplot
plt.subplot(3,1,2)
sns.boxplot(data=df_inv_15 , y='ArrDelay',color=sns.color_palette()[0])
plt.yticks([0,23, 37,69,100,200,300,400,500])
plt.ylim([0,500])
plt.grid(True)

# violinplot
plt.subplot(3,1,3)
sns.violinplot(data=df_inv_15 , y='ArrDelay',color=sns.color_palette()[0],inner="quartile")
plt.yticks([0,23, 37,69,100,200,300,400,500])
plt.ylim([0,500])
plt.grid(True);


# polishing histogram

plt.figure(figsize=(15,5))

sns.histplot(df_inv_15['ArrDelay'],bins=155,kde=True,
             color=sns.color_palette()[0])#,hist_kws={'histtype':'bar','alpha':0.1875}),color='skyblue'
sns.despine()

plt.title('Arrival delay distribution \nfor flights of 15 min. or more delays',
               fontsize=18,color='#BF1B00',weight='bold')

plt.xlabel('Arrival delay (min.)',color='blue',fontsize=15)
plt.xlim([0,500])
plt.xticks([0, 15, 30,46,62,78,93,109,125,141,157,200,215,230,250,300,315,330,350,400,500],
           fontsize='12.5')


plt.ylabel('Freq.',color='blue',fontsize=15);
plt.yticks(np.arange(0,7,1)*10**5,
           [0]+[str(n)+'e5' for n in range(1,7,1)],
           #[0,'1e5','2e5','3e5','4e5','5e5','6e5'],
           fontsize='12.5');


# ArrDelay histograms limited to 150 minutes per Months for all 7+ million original data
sns.set(style='whitegrid')
g = sns.FacetGrid(data=df_inv,col='Month'
                  ,col_wrap=4,
                  xlim=(0,150))
g.fig.suptitle('Arrival Delay (limited to 150 minutes) frequencies per Months\nfor all 7+ million flights data')
g.map(sns.histplot,'ArrDelay',bins=115,)
g.set_titles('{col_name}')
g.set_ylabels('Freq.');#,color='blue',fontsize=15);


# ArrDelay histograms limited to 300 minutes per Months for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')
g = sns.FacetGrid(data=df_inv_15,col='Month'
                  ,col_wrap=4
                  ,xlim=(0,300))
g.fig.suptitle('Arrival Delay (15+ minutes, limited to 300 minutes) frequencies per Months')
g.map(sns.histplot,'ArrDelay',bins=100,)
g.set_titles('{col_name}')
g.set_ylabels('Freq.');#,color='blue',fontsize=15);


# ArrDelay histograms per Months for sampled 500 data points of flights with 15+ minutes arrival delay

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15_samp,col='Month',
                  col_wrap=4,
                  size=3.5,
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes) histograms per Months\nfor sampled 500 flights data')
g.map(sns.histplot,'ArrDelay',bins=20,)
g.set_titles('{col_name}')
g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")


# ArrDelay boxplots per Months for flights with 15+ minutes arrival delay

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15,col='Month',
                  col_wrap=4,
                  size=3.5,
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes) boxplots per Months')
g.map(sns.boxplot,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:643: UserWarning: Using the boxplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)


# ArrDelay limited to 150 min. boxplots per Months for flights with 15+ minutes arrival delay

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15,col='Month',
                  col_wrap=4,
                  size=3.5,xlim=(0,150),
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes,limited to 150 min.) boxplots per Months')
g.map(sns.boxplot,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:643: UserWarning: Using the boxplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)


# ArrDelay boxplots per Months for sampled 500 data points of flights with 15+ minutes arrival delay

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15_samp,col='Month',
                  col_wrap=4,
                  size=3.5,
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes) boxplots per Months\nfor sampled 500 flights data')
g.map(sns.boxplot,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:643: UserWarning: Using the boxplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)


# ArrDelay violinplots per Months for flights with 15+ minutes arrival delay
#colors = (color for color in sns.color_palette()[0:4])
def my_violin(y,*args,**kwargs):
    #color=next(colors)
    sns.violinplot(y=y,)#color=color)#,orient='v')
    
sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15,col='Month',
                  col_wrap=4,
                  size=3.5,
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes) violin plots per Months')
g.map(my_violin,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")


# ArrDelay violinplots limited to 200 min. per Months for flights with 15+ minutes arrival delay

#colors = (color for color in sns.color_palette()[0:4])
def my_violin(y,*args,**kwargs):
    #color=next(colors)
    sns.violinplot(y=y,)#inner="quartile")#color=color)#,orient='v')

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15,col='Month',
                  col_wrap=4,
                  size=3.5,ylim=(0,250),
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes,limited to 200 min.) violin plots per Months')
g.map(my_violin,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")


# ArrDelay violinplots per Months for sampled 500 data points of flights with 15+ minutes arrival delay

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15_samp,col='Month',
                  col_wrap=4,
                  size=3.5,
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes) violin plots per Months\nfor sampled 500 flights data')
g.map(my_violin,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")


# ArrDelay histograms limited to 150 minutes per Carriers for all 7+ million original data
sns.set(style='whitegrid')
g = sns.FacetGrid(data=df_inv,col='Carrier'
                  ,col_wrap=5
                  ,xlim=(0,150))
g.fig.suptitle('Arrival Delay (limited to 150 minutes) frequencies per Carriers\nfor all 7+ million flights data')
g.map(sns.histplot,'ArrDelay',bins=100,)
g.set_titles('{col_name}')
g.set_ylabels('Freq.');#,color='blue',fontsize=15);


# ArrDelay histograms limited to 300 minutes per Carriers for flights with 15+ minutes arrival delay

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15,col='Carrier',
                  col_wrap=5,
                  xlim=(0,300))
g.fig.suptitle('Arrival Delay (15+ minutes, limited to 300 minutes) frequencies per Carriers')
g.map(sns.histplot,'ArrDelay',bins=100,)
g.set_titles('{col_name}')
g.set_ylabels('Freq.');#,color='blue',fontsize=15);


# ArrDelay histograms per Carriers for sampled 500 data points of flights with 15+ minutes arrival delay

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15_samp,col='Carrier',col_order=df_inv_15.Carrier.unique(),
                  col_wrap=5,
                  size=3.5,
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes) histograms per Carriers\nfor sampled 500 flights data')
g.map(sns.histplot,'ArrDelay',bins=20,)
g.set_titles('{col_name}')
g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")


# ArrDelay boxplots per Carriers for flights with 15+ minutes arrival delay

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15,col='Carrier',
                  col_wrap=5,
                  size=3.5,
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes) boxplots per Carriers')
g.map(sns.boxplot,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:643: UserWarning: Using the boxplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)


# ArrDelay (15+ minutes,limited to 200 min.) boxplots per Carriers for flights with 15+ minutes arrival delay

#colors = (color for color in sns.color_palette()[0:4])
def my_box(y,*args,**kwargs):
    #color=next(colors)
    sns.boxplot(y=y,)#color=color)#,orient='v')

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15,col='Carrier',
                  col_wrap=5,
                  size=3.5,ylim=(0,200),
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes,limited to 200 min.) boxplots per Carriers')
g.map(my_box,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")


# ArrDelay boxplots per Carriers for sampled 500 data points of flights with 15+ minutes arrival delay

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15_samp,col='Carrier',col_order=df_inv_15.Carrier.unique(),
                  col_wrap=5,
                  size=3.5,
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes) boxplots per Carriers\nfor sampled 500 flights data')
g.map(sns.boxplot,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:643: UserWarning: Using the boxplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)


# ArrDelay violinplots per Carriers for flights with 15+ minutes arrival delay

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15,col='Carrier',
                  col_wrap=5,
                  size=3.5,
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes) violin plots per Carriers')
g.map(sns.violinplot,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:643: UserWarning: Using the violinplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)


# ArrDelay (15+ minutes,limited to 200 min.) violinplots per Carriers for flights with 15+ minutes arrival delay

#colors = (color for color in sns.color_palette()[0:4])
def my_violin(y,*args,**kwargs):
    #color=next(colors)
    sns.violinplot(y=y,)#color=color)#,orient='v')

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15,col='Carrier',
                  col_wrap=5,
                  size=3.5,ylim=(-20,200),
                  gridspec_kws={'grid':True,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes,limited to 200 min.) violin plots per Carriers')
g.map(my_violin,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")


# ArrDelay violinplots per Carriers for sampled 500 data points of flights with 15+ minutes arrival delay

sns.set(font_scale=1.19,style='whitegrid')
g = sns.FacetGrid(data=df_inv_15_samp,col='Carrier',col_order=df_inv_15.Carrier.unique(),
                  col_wrap=5,
                  size=3.5,
                  gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
g.fig.suptitle('Arrival Delay (15+ minutes) violin plots per Carriers\nfor sampled 500 flights data')
g.map(sns.violinplot,'ArrDelay')#,bins=20,)
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:408: UserWarning: `gridspec_kws` ignored when using `col_wrap`
  warnings.warn("`gridspec_kws` ignored when using `col_wrap`")
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:643: UserWarning: Using the violinplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)


# ArrDelay violinplots per Carriers for sampled 500 data points of flights with 15+ minutes arrival delay 
# the same last plot but with vertical orientation (using catplot)

sns.set(font_scale=1.19,style='whitegrid')
g = sns.catplot(kind = 'violin',data=df_inv_15_samp,col='Carrier',col_order=df_inv_15.Carrier.unique(),col_wrap=5,
                y='ArrDelay',#orient='v',
                size=3.5,
                gridspec_kws={'grid':1,'style':'whitegrid',})
# g = sns.FacetGrid(data=df_inv_15_samp,col='Carrier',col_order=df_inv_15.Carrier.unique(),
#                   col_wrap=5,
#                   size=3.5,
#                   gridspec_kws={'grid':1,'style':'whitegrid',})#'style':'whitegrid',
#sns.set(style='white')
title = 'Arrival Delay (15+ minutes) violin plots per Carriers\nfor sampled 500 flights data'
g.fig.suptitle(title + '\n(with vertical orientation using catplot)',y=1.05)
#g.map(sns.violinplot,y=df_inv_15_samp['ArrDelay'],orient='v')   # vertical orientation
g.set_titles('{col_name}');
#g.set_ylabels('Freq.');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\categorical.py:3747: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)


# finding zero delays or less of delay causes 

delay_causes_cols = ['CarrierDelay','WeatherDelay','NASDelay','SecurityDelay']
causes = ['Carrier','Weather','National Air\nSystem','Security']

print(f"\nFrom {len(df_inv_15):,} flights of 15 minutes arrival delay or more:\n*****") 
for i,col in enumerate(delay_causes_cols):
    outliers = df_inv_15[df_inv_15[col] <=0].shape[0]
    print(f"'{outliers:,}' flights",
          'had "{}" min. or less of {} delay'.format(0,causes[i].replace('\n',' ').lower()))

From 1,524,733 flights of 15 minutes arrival delay or more:
*****
'854,113' flights had "0" min. or less of carrier delay
'1,424,748' flights had "0" min. or less of weather delay
'596,703' flights had "0" min. or less of national air system delay
'1,518,531' flights had "0" min. or less of security delay


# finding causes-delay times outliers
delays = [2500,2000,1500,1000,500,400,350,250]
zero_or_null = [0,np.NaN]

for i,col in enumerate(delay_causes_cols):
    total = df_inv_15[~df_inv_15[col].isin(zero_or_null)]
    print('\nFrom {:,} {}-delay times:\n*****'.format(len(total),causes[i].replace('\n',' ').lower()))
    
    for delay in delays[::-1]:
        outliers = df_inv_15[df_inv_15[col] > delay].shape[0]
        print(f"'{outliers:,}' flights",
              'exceeded "{}" min. {} delay'.format(delay,causes[i].replace('\n',' ').lower()))
        if outliers == 0:
            break

From 670,620 carrier-delay times:
*****
'6,445' flights exceeded "250" min. carrier delay
'2,642' flights exceeded "350" min. carrier delay
'1,837' flights exceeded "400" min. carrier delay
'1,026' flights exceeded "500" min. carrier delay
'116' flights exceeded "1000" min. carrier delay
'6' flights exceeded "1500" min. carrier delay
'1' flights exceeded "2000" min. carrier delay
'0' flights exceeded "2500" min. carrier delay

From 99,985 weather-delay times:
*****
'1,535' flights exceeded "250" min. weather delay
'500' flights exceeded "350" min. weather delay
'315' flights exceeded "400" min. weather delay
'156' flights exceeded "500" min. weather delay
'12' flights exceeded "1000" min. weather delay
'0' flights exceeded "1500" min. weather delay

From 928,030 national air system-delay times:
*****
'3,299' flights exceeded "250" min. national air system delay
'721' flights exceeded "350" min. national air system delay
'359' flights exceeded "400" min. national air system delay
'100' flights exceeded "500" min. national air system delay
'9' flights exceeded "1000" min. national air system delay
'0' flights exceeded "1500" min. national air system delay

From 6,202 security-delay times:
*****
'5' flights exceeded "250" min. security delay
'2' flights exceeded "350" min. security delay
'0' flights exceeded "400" min. security delay


# preparing for analyzing causes delays in df_inv_15  (filtering from 1,524,733 rows)
# looking if there were a lot of zeros or missing values affected the plots

print('Dataset Length: ',df_inv_15.shape[0],'\n')

for delay_col in delay_causes_cols:
    mask = df_inv_15[delay_col].isin(zero_or_null)
    # counting zeros and nulls in delay causes columns 
    print(df_inv_15[mask][delay_col].value_counts(dropna=False))

Dataset Length:  1524733 

0.0    854113
Name: CarrierDelay, dtype: int64
0.0    1424748
Name: WeatherDelay, dtype: int64
0.0    596703
Name: NASDelay, dtype: int64
0.0    1518531
Name: SecurityDelay, dtype: int64


df_inv_15


# preparing data to visualize univariate plots of delay causes frequencies


# preparing df_inv_15  (1,524,733  rows)
id_vars = list(set(df_inv_15.columns)-set(delay_causes_cols))

melt_15 = df_inv_15.melt(id_vars= id_vars,
                         value_vars=delay_causes_cols,
                         var_name='Cause', value_name='Minutes')

# removing zero and missed minutes of delay causes,
# all ArrDelay values already are 15 minutes or more, from previous preparing
zero_or_null = [0,np.NaN]
z_melt_mask = melt_15['Minutes'].isin(zero_or_null)  #--------- 'z_melt_mask' zero and missing mask for melt_15

melt_15 = melt_15[~z_melt_mask].copy()
melt_15.reset_index(drop=True,inplace=True)

# rearranging melt_15 columns
cols = ['Month','Carrier','ArrDelay','Cause','Minutes']
melt_15 = melt_15[cols].copy()

# replacing 'Cause' column values with the causes labels
melt_15['Cause'] = melt_15['Cause'].replace(delay_causes_cols,['Carrier','Weather','National Air System','Security'])

# converting 'Month' to category
melt_15['Month'] = melt_15['Month'].astype(months_cat)

#melt_15.to_csv('2008_melt_15.csv',index=False)
print(melt_15.shape)
melt_15

(1704837, 5)


# melting delay causes columns in flights sample data points 'df_inv_15_samp' ------------ melt_samp_15 (from 'df_inv_15_samp')
melt_samp_15 = df_inv_15_samp.melt(id_vars= id_vars,
                                   value_vars=delay_causes_cols,
                                   var_name='Cause', value_name='Minutes')
# removing zero and missed minutes
# all ArrDelay values already are 15 minutes or more
z_melt_samp_mask = melt_samp_15['Minutes'].isin(zero_or_null)  #----- 'z_melt_samp_mask' zero and missing mask for melt_samp_15 

# converting 'Month' to category
melt_15['Month'] = melt_15['Month'].astype(months_cat)
melt_samp_15 = melt_samp_15[~z_melt_samp_mask].copy()
melt_samp_15.reset_index(drop=True,inplace=True)

# replacing 'Cause' column values with the causes labels
melt_samp_15['Cause'] = melt_samp_15['Cause'].replace(delay_causes_cols,['Carrier','Weather','National Air System','Security'])

# rearranging melt_samp_15 columns
cols = ['Month','Carrier','ArrDelay','Cause','Minutes']
melt_samp_15 = melt_samp_15[cols].copy()
melt_samp_15


# Initial look at causes delays distributions of sample data

# univarite faceted distribution plots of delay causes for sample data points
# limiting to 300 minutes

# histplots
g = sns.FacetGrid(data = melt_samp_15 ,col = 'Cause',
                 xlim=(0,300))

g.map(sns.histplot,'Minutes',bins=20,kde=True)
#g.set_titles('{col_name}')
g.set_titles('')
g.set_ylabels('Freq.',color='blue')

# getting individual delays names for labeling x axis
# xaxis_labels = melt_samp_15['Cause'].unique()
#g.xaxis.set_ticks_position('top');
#g.axes().spines['bottom'].set_position('top')
for i in range(4):
    g.axes[0,i].set_xlabel(causes[i]+' (min.)',color='blue')
    g.axes[0,i].grid(True);


# plotting faceted univariate plots of the previously melted 'delay-causes columns' for sample data (500 rows)
# with limiting up to 150 minutes for delay cause in the analysed sample

# histplots

sns.set_theme(style='ticks')

g = sns.FacetGrid(data = melt_samp_15 ,
                  col = 'Cause',hue='Cause',palette='bright',
                  xlim=(0,150),
                  height=4,aspect=.77)
sns.set(font_scale=1.19)             
g.fig.suptitle('Delay causes distributions for sampled flights data points \nof 15 min. or more arrival delays',
               fontsize=18,color='#BF1B00',weight='bold')

g.map(sns.histplot,'Minutes',bins=35,kde=True,)
#g.set_titles('{col_name}')

g.set_titles('')

xticks = [0,12,25,37,50,62,75,88,100,125,150]
g.set(xticks=xticks)
#g.fig.xticklabels(xticks,fontsize=12.5)
#g.axes[0,i].set_xticklabels(xticks,color='blue',fontsize=12.5)

#g.xaxis.set_ticks_position('top');
#g.axes().spines['bottom'].set_position('top')

g.set_ylabels('Freq.',color='blue',fontsize=18)

for i in range(4):
    g.axes[0,i].set_xlabel(causes[i]+' delay (min.)',fontsize=15,color=sns.color_palette('bright')[i])
    g.axes[0,i].grid(True);


# plotting faceted univariate boxplots of the previously melted 'delay-causes columns' for sample data (500 rows)

# creating my_box function for coloring different causes-delays boxplots, and adjusting vertical orientation
colors = (color for color in sns.color_palette()[0:4])
def my_box(y,color=None,*args,**kwargs):
    color=next(colors)
    sns.boxplot(y=y,color=color,*args,**kwargs)#,orient='v')

# plotting
g = sns.FacetGrid(data = melt_samp_15 ,
                  col = 'Cause',
                  #ylim=(-15,100),
                  height=3.5,aspect=1)
#g.map(sns.boxplot,'Minutes')
g.map(my_box,'Minutes')

sns.set(font_scale=1.19,style='whitegrid')
sns.despine(bottom=True)
#g.set_titles('{col_name}')
g.set_titles('')
# yticks = [-15,0,10,20,35,50,70,100]
# g.set(yticks=yticks)

#g.xaxis.set_ticks_position('top');
#g.axes().spines['bottom'].set_position('top')
for i in range(4):
    g.axes[0,i].set_ylabel(delay_causes_cols[i]+' (min.)')
    #g.axes[0,i].grid(True);

g.set_xlabels('');
g.fig.subplots_adjust(wspace=.2);                       # adjusing faceted plots width spacing
#.set_ylabels('Est. Distributions');


# plotting faceted univariate boxplots of the previously melted 'delay-causes columns' for sample data (500 rows)
# with limiting up to 100 minutes for delay cause in the analysed sample

# creating my_box function for coloring different causes-delays boxplots, and adjusting vertical orientation
colors = (color for color in sns.color_palette()[0:4])
def my_box(y,color=None,*args,**kwargs):
    color=next(colors)
    sns.boxplot(y=y,color=color,*args,**kwargs)#,orient='v')

# plotting
g = sns.FacetGrid(data = melt_samp_15 ,
                  col = 'Cause',
                  ylim=(-15,100),
                  height=3.5,aspect=1)
#g.map(sns.boxplot,'Minutes')
g.map(my_box,'Minutes')

sns.set(font_scale=1.19,style='whitegrid')
sns.despine(bottom=True)
#g.set_titles('{col_name}')
g.set_titles('')
yticks = [-15,0,10,20,35,50,70,100]
g.set(yticks=yticks)

#g.xaxis.set_ticks_position('top');
#g.axes().spines['bottom'].set_position('top')
for i in range(4):
    g.axes[0,i].set_ylabel(delay_causes_cols[i]+' (min.)')
    #g.axes[0,i].grid(True);

g.set_xlabels('');
g.fig.subplots_adjust(wspace=.2);                       # adjusing faceted plots width spacing
#.set_ylabels('Est. Distributions');


# faceted univariate violinplots of the previously melted 'delay-causes columns' for sample data (500 rows)

# creating my_violin function for coloring different causes-delays violinplots, and adjusting vertical orientation
colors = (color for color in sns.color_palette()[0:4])
def my_violin(y,color=None,*args,**kwargs):
    color=next(colors)
    sns.violinplot(y=y,color=color)#,orient='v')

# plotting
g = sns.FacetGrid(data = melt_samp_15,
                  col = 'Cause',hue='Cause',#row='Month',
                  height=4.5,aspect=.9)

title = 'Causes-delays distributions \nfor the year 2008 flights sampled data points \nof 15 min. or more arrival delays'
g.fig.suptitle(title,
               fontsize=21,
               color='#BF1B00',
               weight='bold',y=1.04)

sns.despine(left=True,bottom=True)
sns.set(style='whitegrid')

g.map(sns.violinplot,'Minutes')#,orient='v')

# g.set_titles('{col_name}')

g.set_titles('')
xticks = [-60,0,100,200,300,400]
g.set(xticks=xticks)
g.set_xticklabels(xticks,fontsize=13.5)

g.set_ylabels('');

for i in range(4):
    g.axes[0,i].set_xlabel(causes[i]+' delay (min.)',fontsize=18,color='blue')
    #g.axes[0,i].grid(True);
#g.set_axis_labels(x_var='causes', y_var='None')
#g.set_ylabels('Est. Distributions');
    
#g.xaxis.set_ticks_position('top')
#g.axes().spines['bottom'].set_position('top')

#g.set_grid(True,axis='y');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:643: UserWarning: Using the violinplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)


# faceted univariate violinplots of the previously melted 'delay-causes columns' for sample data (500 rows)

# creating my_violin function for coloring different causes-delays violinplots, and adjusting vertical orientation
colors = (color for color in sns.color_palette()[0:4])
def my_violin(y,color=None,*args,**kwargs):
    color=next(colors)
    sns.violinplot(y=y,color=color)#,orient='v')

# plotting
g = sns.FacetGrid(data = melt_samp_15,
                  col = 'Cause',col_wrap=2,
                  hue='Cause',#row='Month',
                  height=3.5,aspect=1.5)

title = 'Causes-delays (limited to 150) distributions \nfor the year 2008 flights sampled data points'
g.fig.suptitle(title + '\nof 15+ min. arrival delays',
               fontsize=21,
               color='#BF1B00',
               weight='bold')#,y=1.04)

sns.despine(left=True,bottom=True)
sns.set(style='whitegrid')#,font_scale=1.19,

g.map(sns.violinplot,'Minutes',inner='quartile')#,orient='v')

# g.set_titles('{col_name}')

g.set_titles('')
xticks = [-40,0,18,26,36,65,78,115,150]#,200,250]
g.set(xticks=xticks,xlim=(-40,150))
g.set_xticklabels(xticks,fontsize=14)

#g.set_xlabels('');

g.fig.subplots_adjust(wspace=.2)

xlabel = (label for label in causes)
colors = (color for color in sns.color_palette()[0:4])
for i in range(4):
    g.axes[i].set_xlabel(next(xlabel)+' delay (min.)',fontsize=18,color=next(colors));
    #g.axes[1,i].set_xlabel(causes[i]+' delay (min.)',fontsize=18,color=next(colors))
    #g.axes[0,i].grid(True);
#g.set_axis_labels(x_var='causes', y_var='None')
#g.set_ylabels('Est. Distributions');
    
#g.xaxis.set_ticks_position('top')
#g.axes().spines['bottom'].set_position('top')

#g.set_grid(True,axis='y');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:643: UserWarning: Using the violinplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)


# plotting delay causes distributions in df_inv_15 without zero or missing values (filtering from 1,524,733 rows)
# limiting max minutes to 3000 to avoid  a number of outliers

# histograms
plt.figure(figsize=(20,5.95))

sns.set(style='ticks')
plt.suptitle('Causes-delays distributions for flights of 15 min. or more arrival delays',
               fontsize=21,color='#BF1B00',weight='bold')

for i,delay_col in enumerate(delay_causes_cols):
    
    mask = df_inv_15[delay_col].isin(zero_or_null)
    cause_col = df_inv_15[~mask][delay_col]
    
    plt.subplot(1,4,i+1)
    sns.histplot(cause_col,bins=155,kde=True,color=sns.color_palette('dark')[i])
    
    sns.despine()
    #plt.xlim((0,3800))
    #xtick = [0,300,700,1300,1800,2400,3000,3800]
    #plt.xticks(xtick,xtick,fontsize=13.5)
    
    #plt.ylim((0,850))
    #ytick = [0,450,850]
    #plt.yticks(ytick,ytick,fontsize=13.5)

    #plt.yticks(np.arange(0,1100,250),fontsize=12.5)
    
    plt.xlabel('\n'+causes[i]+' delay (min.)',color=sns.color_palette('dark')[i],fontsize=16)
    plt.ylabel('Freq.',color=sns.color_palette('dark')[i],fontsize=16)#,color='blue'
    plt.grid(True)

plt.subplots_adjust(wspace=.2) #################################################


# plotting delay causes distributions in df_inv_15 without zero or missing values (filtering from 1,524,733 rows)
# limiting max minutes to 3000 to avoid  a number of outliers

# histograms
plt.figure(figsize=(20,10))

sns.set(style='ticks')

title='Causes-delays distributions \nfor the year 2008 flights of 15+ min. arrival delays'
plt.suptitle(title+'\n\n(independent axes)',y=1.04,
             fontsize=21,color='black',# '#BF1B00',
             weight='bold')

colors = (color for color in sns.color_palette('dark'))    

for i,delay_col in enumerate(delay_causes_cols):
    
    mask = df_inv_15[delay_col].isin(zero_or_null)
    cause_col = df_inv_15[~mask][delay_col]
    
    plt.subplot(2,2,i+1)
    
    color = next(colors)
    ax = sns.histplot(cause_col,bins=155,kde=False,color = color)
    
    sns.despine()
    
    plt.xlabel('\n'+causes[i]+' delay (min.)',color=color,fontsize=18,weight='bold')
    plt.ylabel('Freq.',color=color,fontsize=18,weight='bold')
    
    #plt.xlim((0,2500))
    #xtick = [0,300,700,1300,1800,2400,3000,3800]
    #plt.xticks(xtick,xtick,fontsize=13.5)
    plt.xticks(fontsize=14,weight='bold',color=color)
    
    #plt.ylim((0,300000))
    #ytick = [0,450,850]
    #plt.yticks(ytick,ytick,fontsize=13.5)
    
    plt.yticks(fontsize=14,weight='bold',color=color)
    
    # adjusting yticklabels
    if i == 0:
        yticks = (ax.get_yticks()).astype('int')
        yticklabels = [0] + [str(n/100000)+'e5' for n in yticks[1:]]
        plt.yticks(ticks=yticks,
                   labels=yticklabels,
                   fontsize=14,weight='bold',color=color)
    elif i == 1:
        yticks = (ax.get_yticks()).astype('int')
        yticklabels = [0] + [str(n/1000)+'e3' for n in yticks[1:9]] +['']
        plt.yticks(ticks=yticks,
                   labels=yticklabels,
                   fontsize=14,weight='bold',color=color)
    elif i == 2:
        yticks = (ax.get_yticks()).astype('int')
        yticklabels = [0] + [str(n/100000)+'e5' for n in yticks[1:6]] +['']
        plt.yticks(ticks=yticks,
                   labels=yticklabels,
                   fontsize=14,weight='bold',color=color)
        
    plt.grid(True,axis='y')

plt.subplots_adjust(hspace=.4,wspace=.3)


# polished histogram

sns.set_theme(style='ticks')
plt.figure(figsize=(15,16))

plt.suptitle('Causes delays distribution \nfor the year 2008 flights of 15 min. or more delays \n (limiting to 500 min.)',
               fontsize=21,color='#BF1B00',weight='bold',y=.94)   
        
for i,delay_col in enumerate(delay_causes_cols):
    
    mask = df_inv_15[delay_col].isin(zero_or_null)
    cause_col = df_inv_15[~mask][delay_col]
    
    plt.subplot(4,1,i+1)
    sns.histplot(cause_col,bins=70,kde=True,
                 color=sns.color_palette()[9])#,hist_kws={'histtype':'bar','alpha':0.1875}),color='skyblue'
    sns.despine()
    plt.xlabel(causes[i]+' delay (min.)',color='blue',fontsize=15)
    plt.ylabel('Freq.',color='blue',fontsize=15);
    plt.xlim([0,500])
    plt.xticks([0, 20, 37,46,62,78,93,109,125,141,157,200,215,230,250,300,315,330,350,400,500],
               fontsize='12.5')
    plt.ylim((0,700000))
    plt.yticks(np.arange(0,7,1)*10**5,
               [0]+[str(n)+'e5' for n in range(1,7,1)],
               #[0,'1e5','2e5','3e5','4e5','5e5','6e5'],
               fontsize=13,weight='bold')
plt.subplots_adjust(hspace=.4);


# plotting boxplots delay causes distributions in df_inv_15 without zero or missing values (filtering from 1,524,733 rows)
# with shared 2500 minutes axis

# boxplots
plt.figure(figsize=(15,6))

sns.set(style='ticks')

for i,delay_col in enumerate(delay_causes_cols):
    
    mask = df_inv_15[delay_col].isin(zero_or_null)
    cause_col = df_inv_15[~mask][delay_col]
    
    plt.subplot(1,4,i+1)
    sns.boxplot(y=cause_col,color=sns.color_palette('muted')[i])
    
    sns.despine(bottom=True)
    #ytick = [0,200,500,1000,1500]
    #plt.yticks(xtick,xtick)
    plt.ylim((0,2500))
    plt.ylabel(delay_col + ' (min.)',color=sns.color_palette('muted')[i],weight='bold')
    #plt.ylabel('Est. Distributions')
plt.subplots_adjust(wspace=.4)


# plotting boxplots delay causes distributions in df_inv_15 without zero or missing values (filtering from 1,524,733 rows)
# limiting axis to 150 min.

# boxplots
plt.figure(figsize=(15,6))

sns.set(style='ticks')

for i,delay_col in enumerate(delay_causes_cols):
    
    mask = df_inv_15[delay_col].isin(zero_or_null)
    cause_col = df_inv_15[~mask][delay_col]
    
    plt.subplot(1,4,i+1)
    sns.boxplot(y=cause_col,color=sns.color_palette('muted')[i])
    
    sns.despine(bottom=True)
    
    ytick = [0,7,13,18,25,41,57,65,89,126,150]
    plt.yticks(ytick,ytick,fontsize=13.25)
    plt.ylim((0,150))
    plt.ylabel(delay_col + ' (min.)',color=sns.color_palette('muted')[i],weight='bold',fontsize=15)
    #plt.ylabel('Est. Distributions')
    plt.grid()
plt.subplots_adjust(wspace=.4)


# plotting violinplots delay causes distributions in df_inv_15 without zero or missing causes-delays values
# (filtering from 1,524,733 rows)
# limiting max minutes to 2500

# violinplots
plt.figure(figsize=(15,6))

sns.set(style='ticks')

for i,delay_col in enumerate(delay_causes_cols):
    
    mask = df_inv_15[delay_col].isin(zero_or_null)
    cause_col = df_inv_15[~mask][delay_col]
    
    plt.subplot(1,4,i+1)
    sns.violinplot(y=cause_col,color=sns.color_palette('muted')[i])#,palette='deep'
    sns.despine(bottom=True)
    
    ytick = [-50,0,200,500,1000,1500,2500]
    plt.yticks(ytick,ytick)
    plt.ylim((-15,2500))
    plt.ylabel(delay_col+' (min.)',color=sns.color_palette('muted')[i],weight='bold')
    plt.grid(True)
    #plt.ylabel('Est. Distributions')
plt.subplots_adjust(wspace=.5)


# plotting violinplots delay causes distributions in df_inv_15 without zero or missing causes-delays values
# (filtering from 1,524,733 rows)


# violinplots
plt.figure(figsize=(15,6))

sns.set(style='ticks')

for i,delay_col in enumerate(delay_causes_cols):
    
    mask = df_inv_15[delay_col].isin(zero_or_null)
    cause_col = df_inv_15[~mask][delay_col]
    
    plt.subplot(1,4,i+1)
    sns.violinplot(y=cause_col,color=sns.color_palette()[i])#,palette='deep'
    sns.despine(bottom=True)
    
    ytick = [-15,0,13,18,26,36,65,78,115,150,200,250]
    plt.yticks(ytick,ytick)
    plt.ylim((-15,150))
    plt.ylabel(delay_col+' (min.)',color=sns.color_palette()[i],weight='bold')
    plt.grid(True)
    #plt.ylabel('Est. Distributions')
plt.subplots_adjust(wspace=.5)


# delay-causes delays histograms limited to 250 minutes per Months for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_15,col='Cause',col_order=order,
                  row='Month',margin_titles=True,
                  hue='Cause',hue_order=order,
                  #col_wrap=4,
                  xlim=(0,250),
                  palette='muted',
                 )#size=4)#height=5,aspect=.77)
sns.set(font_scale=1.19)
title = 'Causes-Delays (limited to 250 min.) frequencies per Months\n(for flights with 15+ minutes arrival delay)'
g.fig.suptitle(title,x=.331,y=1.015)
g.map(sns.histplot,'Minutes',element ='step',fill=True,alpha=.185,bins=70)
g.set_titles(col_template='{col_name}',row_template='{row_name}')
g.add_legend(title = 'Delay causes',loc='upper center',ncol=4,bbox_to_anchor=(0.331,1.005))
g.set_xlabels('Delay (min.)')
g.set_ylabels('Freq.');#,color='blue',fontsize=15);


# delay-causes delays boxplots per Months for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')

order = melt_15.Cause.unique()
colors = [color for color in sns.color_palette('muted')[0:4]] * 12
colors = (color for color in colors)

def my_box_2(y,*args,**kwargs):
    color=next(colors)
    sns.boxplot(y=y,color=color)
    
# g = sns.catplot(kind='box',
#                 data=melt_15,col='Cause',col_order=order,
#                 row='Month',margin_titles=True,
#                 order=order,
#                 hue='Cause',hue_order=order,
#                 y='Minutes',
#                 #col_wrap=4,
#                 #xlim=(0,250),
#                 palette='muted',
#                 size=3.5)#height=5,aspect=.77)

g = sns.FacetGrid(#kind='box',color=next(colors),
                  data=melt_15,col='Cause',col_order=order,
                  row='Month',margin_titles=True,
                  #hue='Cause',hue_order=order,#hue_kws={'color': [color for color in list(sns.color_palette('muted')[0:4])]},
                  #y='Minutes',
                  #col_wrap=4,
                  palette='muted',
                  size=3.5)#height=5,aspect=.77)

sns.set(font_scale=1.19)
title = 'Causes-Delays boxplots per Months\n(for flights with 15+ minutes arrival delay)'
g.fig.suptitle(title,y=1)#,x=.5,y=1.015)
g.map(my_box_2,'Minutes')#,orient='v')
g.set_titles(col_template='{col_name}',row_template='{row_name}')
#g.add_legend(title = 'Delay causes',loc='upper center',ncol=4,bbox_to_anchor=(0.331,1.005))
g.set_ylabels('Delay (min.)')
g.set_xlabels('');
#g.set_ylabels('Freq.');#,color='blue',fontsize=15);

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)


# delay-causes delays boxplots limited to 150 minutes per Months for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')

order = melt_15.Cause.unique()
colors = [color for color in sns.color_palette('muted')[0:4]] * 12
colors = (color for color in colors)
    
# g = sns.catplot(kind='box',color=next(colors),
#                 data=melt_15,col='Cause',col_order=order,
#                 row='Month',margin_titles=True,
#                 hue='Cause',hue_order=order,#hue_kws={'color': [color for color in list(sns.color_palette('muted')[0:4])]},
#                 y='Minutes',
#                 #col_wrap=4,
#                 palette='muted',
#                 size=3.5)#height=5,aspect=.77)

g = sns.FacetGrid(#kind='box',color=next(colors),
                  data=melt_15,col='Cause',col_order=order,
                  row='Month',margin_titles=True,
                  #hue='Cause',hue_order=order,#hue_kws={'color': [color for color in list(sns.color_palette('muted')[0:4])]},
                  #y='Minutes',
                  #col_wrap=4,
                  palette='muted',
                  size=3.5)#height=5,aspect=.77)

sns.set(font_scale=1.19)
g.set(ylim=(0,150),)
title = 'Causes-Delays (limited to 150 min.) boxplots per Months\n(for flights with 15+ minutes arrival delay)'
g.fig.suptitle(title,y=1)#.015)#x=.5,)
g.map(my_box_2,'Minutes')#,*args,**kwargs)#,order=order,orient='v')
g.set_titles(col_template='{col_name}',row_template='{row_name}')
#g.add_legend(title = 'Delay causes',loc='upper center',ncol=4,bbox_to_anchor=(0.5,1.005))#0.331
g.set_ylabels('Delay (min.)')
g.set_xlabels('');
#g.set_ylabels('Freq.');#,color='blue',fontsize=15);

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)


# delay-causes delays boxplots per Months for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')

order = melt_15.Cause.unique()
colors = [color for color in sns.color_palette('muted')[0:4]] * 12
colors = (color for color in colors)

def my_violin_2(y,*args,**kwargs):
    color=next(colors)
    sns.violinplot(y=y,color=color)

# g = sns.catplot(kind='violin',
#                 data=melt_15,col='Cause',col_order=order,
#                 row='Month',margin_titles=True,
#                 order=order,
#                 hue='Cause',hue_order=order,
#                 y='Minutes',
#                 #col_wrap=4,
#                 #xlim=(0,250),
#                 palette='muted',
#                 size=3.5)#height=5,aspect=.77)

g = sns.FacetGrid(#kind='box',color=next(colors),
                  data=melt_15,col='Cause',col_order=order,
                  row='Month',margin_titles=True,
                  #hue='Cause',hue_order=order,#hue_kws={'color': [color for color in list(sns.color_palette('muted')[0:4])]},
                  #y='Minutes',
                  #col_wrap=4,
                  palette='muted',
                  size=3.5)#height=5,aspect=.77)

sns.set(font_scale=1.19)
title = 'Causes-Delays violinplots per Months\n(for flights with 15+ minutes arrival delay)'
g.fig.suptitle(title,y=1)#,x=.5,y=1.015)
g.map(my_violin_2,'Minutes')#,orient='v')
g.set_titles(col_template='{col_name}',row_template='{row_name}')
#g.add_legend(title = 'Delay causes',loc='upper center',ncol=4,bbox_to_anchor=(0.331,1.005))
g.set_ylabels('Delay (min.)')
g.set_xlabels('');
#g.set_ylabels('Freq.');#,color='blue',fontsize=15);

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)


# delay-causes delays boxplots limited to 250 minutes per Months for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')

order = melt_15.Cause.unique()
colors = [color for color in sns.color_palette('muted')[0:4]] * 12
colors = (color for color in colors)

# g = sns.catplot(kind='violin',color=next(colors),
#                 data=melt_15,col='Cause',col_order=order,
#                 row='Month',margin_titles=True,
#                 hue='Cause',hue_order=order,#hue_kws={'color': [color for color in list(sns.color_palette('muted')[0:4])]},
#                 y='Minutes',
#                 #col_wrap=4,
#                 palette='muted',
#                 size=3.5)#height=5,aspect=.77)

g = sns.FacetGrid(#kind='box',color=next(colors),
                  data=melt_15,col='Cause',col_order=order,
                  row='Month',margin_titles=True,
                  #hue='Cause',hue_order=order,#hue_kws={'color': [color for color in list(sns.color_palette('muted')[0:4])]},
                  #y='Minutes',
                  #col_wrap=4,
                  palette='muted',
                  size=3.5)#height=5,aspect=.77)

sns.set(font_scale=1.19)
g.set(ylim=(-20,250),)
title = 'Causes-Delays (limited to 250 min.) violinplots per Months\n(for flights with 15+ minutes arrival delay)'
g.fig.suptitle(title,y=1)#,x=.5,y=1.015)
g.map(my_violin_2,'Minutes')#,color=next(colors))#,order=order,orient='v')
g.set_titles(col_template='{col_name}',row_template='{row_name}')
#g.add_legend(title = 'Delay causes',loc='upper center',ncol=4,bbox_to_anchor=(0.5,1.005))#0.331
g.set_ylabels('Delay (min.)')
g.set_xlabels('');
#g.set_ylabels('Freq.');#,color='blue',fontsize=15);

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:316: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)


# delay-causes delays bivariate histogram for flights with 15+ minutes arrival delay
sns.histplot(data=melt_15,x='Minutes',hue='Cause',bins=155)
plt.xlim((0,250))
plt.title('Causes-delays frequencies (limited to 250)')
plt.xlabel('Delay (min.)')
plt.ylabel('Freq.');


# plotting bivariate boxplots of the previously melted 'delay-causes columns' for sample data (500 rows)

sns.set_style('white')
plt.figure(figsize=(15,8))

plt.subplot(121)

sns.set(font_scale=1.19,style='ticks')

# main plot
ax = sns.boxplot(data= melt_samp_15,
                 y='Minutes',x='Cause',palette='deep')
plt.title('Causes-delays boxplots\nfor sampled data points \nof 15 min. or more delays',
          fontsize=21,
          color='#BF1B00',
          weight='bold')
#yticks=[0,15,35,45,60,100,150,200,250,400]
#plt.yticks(yticks,yticks)

#ax.set_yticks([0,35,45,50,60,100,150,200,250,400])
ax.set_ylim([0,400])
ax.set_xticklabels(causes)
plt.ylabel('Delay (min.)',fontsize=18,color='blue')
plt.xlabel('Cause',fontsize=18,color='blue')
yticks=[0,18,26,36,65,78,115,150,200,250,400]

plt.yticks(yticks,yticks,fontsize=13)
plt.xticks(fontsize=14)
#ax.set_yticklabels([0,30,33,35,36,40,50,60,100,150,200,250,400])

#leg = ax.legend(['Carrier','Weather','National Air System','Security'],
#                loc=1,bbox_to_anchor =(1.7,-0.1),ncol=4)
#for line in leg.get_lines():
#    line.set_linewidth(7.0)
#   lgd.set_color('black') 

sns.despine(left=True)
ax.grid(True,axis ='y')

# second plot for focusing on IQR rectangles and limiting arrival delay axis to 150 minutes 
plt.subplot(122)
sns.set(font_scale=1.19,style='white')

ax = sns.boxplot(data= melt_samp_15,
                 y='Minutes',x='Cause',palette='deep')

sns.despine(bottom=True)
plt.yticks(fontsize=18)
ax.set_yticks([0,18,27,34,43,65,78,115,150,200,250])
ax.set_ylim([0,150])

ax.grid(True,axis ='y');
ax.set_xticks([])
plt.xlabel('Highlighting IQR rectangles \non the left boxplots',fontsize=21,color='Brown')
plt.ylabel('')
plt.subplots_adjust(wspace=.2);


# plotting bivariate boxplots of the previously melted 'delay-causes columns' for sample data (500 rows)

plt.figure(figsize=(15,8))

plt.subplot(121)
sns.set(font_scale=1.19,style='ticks')

# main plot
ax = sns.boxplot(data= melt_15,
                 y='Minutes',x='Cause',palette='pastel')
plt.title('Causes-delays boxplots \nfor all data points \nof 15 min. or more delays',
          fontsize=21,
          color='#BF1B00',
          weight='bold')

plt.ylabel('Delay (min.)',fontsize=18,color='blue')
plt.xlabel('Cause',fontsize=18,color='blue')

ax.set_xticklabels(causes)
plt.xticks(fontsize=14)

#ax.set_ylim([0,400])
#yticks=[0,18,26,36,65,78,115,150,200,250,400]
#plt.yticks(yticks,yticks,fontsize=13)

#ax.set_yticks([0,35,45,50,60,100,150,200,250,400])
#ax.set_yticklabels([0,30,33,35,36,40,50,60,100,150,200,250,400])

#leg = ax.legend(['Carrier','Weather','National Air System','Security'],
#                loc=1,bbox_to_anchor =(1.7,-0.1),ncol=4)
#for line in leg.get_lines():
#    line.set_linewidth(7.0)
#   lgd.set_color('black') 

sns.despine(left=True)
ax.grid(True,axis ='y')

# second plot for focusing on IQR rectangles and limiting arrival delay axis to 150 minutes 
plt.subplot(122)
sns.set(font_scale=1.19,style='white')

ax = sns.boxplot(data= melt_15,
                 y='Minutes',x='Cause',palette='pastel')

sns.despine(bottom=True)
plt.yticks(fontsize=18)
ax.set_yticks([0,7,13,18,25,41,57,65,89,126,150])
ax.set_ylim([0,150])

ax.grid(True,axis ='y');
ax.set_xticks([])
plt.xlabel('Highlighting IQR rectangles \non the left boxplots',fontsize=21,color='Brown')
plt.ylabel('')
plt.subplots_adjust(wspace=.2);


# plotting bivariate boxplots of the previously melted 'delay-causes columns' for sample data (500 rows)

plt.figure(figsize=(15,8))

plt.subplot(121)
sns.set(font_scale=1.19,style='ticks')

# main plot
ax = sns.violinplot(data= melt_samp_15,
                 y='Minutes',x='Cause',palette='deep')
plt.title('Delays-per-causes violinplots\nfor sampled data points \nof 15 min. or more delays',
          fontsize=21,
          color='#BF1B00',
          weight='bold')

plt.ylabel('Delay (min.)',fontsize=18,color='blue')
plt.xlabel('Cause',fontsize=18,color='blue')

ax.set_xticklabels(causes)
plt.xticks(fontsize=14)


#ax.set_ylim([-40,400])
#yticks=[-40,0,18,26,36,65,78,115,150,200,250,400]
#plt.yticks(yticks,yticks,fontsize=13)

#ax.set_yticks([0,35,45,50,60,100,150,200,250,400])
#ax.set_yticklabels([0,30,33,35,36,40,50,60,100,150,200,250,400])

#leg = ax.legend(['Carrier','Weather','National Air System','Security'],
#                loc=1,bbox_to_anchor =(1.7,-0.1),ncol=4)
#for line in leg.get_lines():
#    line.set_linewidth(7.0)
#   lgd.set_color('black') 

sns.despine(left=True)
ax.grid(True,axis ='y')

# second plot for focusing on IQR rectangles and limiting arrival delay axis to 150 minutes 
plt.subplot(122)
sns.set(font_scale=1.19,style='white')

ax = sns.violinplot(data= melt_samp_15,
                 y='Minutes',x='Cause',palette='deep')

sns.despine(bottom=True)
plt.yticks(fontsize=18)
ax.set_yticks([-40,0,18,26,36,65,78,115,150,200,250])
ax.set_ylim([-40,150])

ax.grid(True,axis ='y');
ax.set_xticks([])
plt.xlabel('Highlighting the colored distribution \nof the left violinplots',fontsize=21,color='Brown')
plt.ylabel('')
plt.subplots_adjust(wspace=.2);


# plotting univariate boxplots of the previously melted 'delay-causes columns' for sample data (500 rows)

plt.figure(figsize=(15,8))

plt.subplot(121)
sns.set(font_scale=1.19,style='ticks')

# main plot
ax = sns.violinplot(data= melt_15,
                 y='Minutes',x='Cause',palette='deep')

plt.title('Causes-Delays violinplots\nfor the year 2008 flights \nof 15 min. or more delays',
          fontsize=21,
          color='#BF1B00',
          weight='bold',y=1.04)

plt.xlabel('Cause',fontsize=18,color='blue')
plt.ylabel('Delay (min.)',fontsize=18,color='blue')

ax.set_xticklabels(causes)
plt.xticks(fontsize=14)

#ax.set_ylim([-40,400])
#yticks=[-40,0,18,26,36,65,78,115,150,200,250,400]
#plt.yticks(yticks,yticks,fontsize=13)

sns.despine(left=True)
ax.grid(True,axis ='y')

#ax.set_yticks([0,35,45,50,60,100,150,200,250,400])
#ax.set_yticklabels([0,30,33,35,36,40,50,60,100,150,200,250,400])

#leg = ax.legend(['Carrier','Weather','National Air System','Security'],
#                loc=1,bbox_to_anchor =(1.7,-0.1),ncol=4)
#for line in leg.get_lines():
#    line.set_linewidth(7.0)
#   lgd.set_color('black') 


# second plot for focusing on IQR rectangles and limiting arrival delay axis to 150 minutes 
plt.subplot(122)
sns.set(font_scale=1.19,style='white')

ax = sns.violinplot(data= melt_15,
                 y='Minutes',x='Cause',palette='deep')

sns.despine(bottom=True)
plt.yticks(fontsize=18)
ax.set_yticks([-15,0,13,18,26,36,65,78,115,150,200,250])
ax.set_ylim([-13,150])

ax.grid(True,axis ='y');
ax.set_xticks([])
plt.title('Highlighting the colored distribution \non the left violinplots',fontsize=21,color='#AD1FFF')
plt.ylabel('')
plt.subplots_adjust(wspace=.2);


# delay-causes delays bivariate histograms limited to 250 minutes per Months for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_15,col='Month',
                  hue='Cause',#hue_order=order,
                  col_wrap=4,
                  xlim=(0,250),
                  palette='muted')

g.fig.suptitle('Causes-Delays (limited to 250 min.) frequencies per Months\n(for flights with 15+ minutes arrival delay)')
g.map(sns.histplot,'Minutes',element ='step',fill=True,alpha=.185,bins=70)
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xlabels('Delay (min.)')
g.set_ylabels('Freq.');#,color='blue',fontsize=15);


# delay-causes delays bivariate histograms per Months for sampled 
# 500 data points of flights with 15+ minutes arrival delay

sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_samp_15,col='Month',
                  hue='Cause',#hue_order=order,
                  col_wrap=4,
                  #xlim=(0,250),
                  palette='muted')
title = 'Causes-Delays frequencies per Months\n(for sampled 500 flights with 15+ minutes arrival delay)'
g.fig.suptitle(title)
g.map(sns.histplot,'Minutes',element ='step',fill=True,alpha=.185,bins=18)
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xlabels('Delay (min.)')
g.set_ylabels('Freq.');#,color='blue',fontsize=15);


# delay-causes delays boxplots per Months for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_15,col='Month',
                  
                  col_wrap=4,
                  #xlim=(0,250),
                  palette='muted')

g.fig.suptitle('Causes-Delays boxplots per Months\n(for flights with 15+ minutes arrival delay)')
g.map(sns.boxplot,'Minutes','Cause',order=order,hue=melt_15['Cause'],hue_order=order,orient='h')
g.set_titles('{col_name}')
g.set_xlabels('Delay (min.)')
g.add_legend(title = 'Delay causes');
#g.set_ylabels('Freq.');#,color='blue',fontsize=15);


# delay-causes delays boxplots per Months for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_15,col='Month',
                  
                  col_wrap=4,
                  ylim=(0,100),
                 )#palette='muted')

g.fig.suptitle('Causes-Delays (limited to 100 min.) boxplots per Months\n(for flights with 15+ minutes arrival delay)')
g.map(sns.boxplot,'Cause','Minutes',order=order,hue=melt_15['Cause'],hue_order=order,orient='v')#,palette='rocket_r')
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xticklabels(rotation = 90)
g.set_ylabels('Delay (min.)');#,color='blue',fontsize=15);


# Causes-delays violinplots per Months for flights with 15+ minutes arrival delay

sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_15,col='Month',
                  
                  col_wrap=4,
                  #ylim=(0,100),
                 )#palette='muted')

g.fig.suptitle('Causes-Delays violinplots per Months\n(for flights with 15+ minutes arrival delay)')
g.map(sns.violinplot,'Cause','Minutes',order=order,hue=melt_15['Cause'],hue_order=order,orient='v')#,palette='rocket_r')
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xticklabels(rotation = 90)
g.set_ylabels('Delay (min.)');#,color='blue',fontsize=15);


# Causes-delays (limited to 100 min.) violinplots per Months for flights with 15+ minutes arrival delay

sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_15,col='Month',
                  
                  col_wrap=4,
                  ylim=(-20,100),
                 )#palette='muted')

g.fig.suptitle('Causes-Delays (limited to 100 min.) violinplots per Months\n(for flights with 15+ minutes arrival delay)')
g.map(sns.violinplot,'Cause','Minutes',order=order,hue=melt_15['Cause'],hue_order=order,orient='v')#,palette='rocket_r')
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xticklabels(rotation = 90)
g.set_ylabels('Delay (min.)');#,color='blue',fontsize=15);


# Causes-delays violinplots per Months for sampled
# 500 flights with 15+ minutes arrival delay

sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_samp_15,col='Month',
                  
                  col_wrap=4,
                  #ylim=(-20,100),
                 )#palette='muted')

g.fig.suptitle('Causes-Delays violinplots per Months\n(for sampled 500 flights with 15+ minutes arrival delay)')
g.map(sns.violinplot,'Cause','Minutes',order=order,hue=melt_15['Cause'],hue_order=order,orient='v')#,palette='rocket_r')
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xticklabels(rotation = 90)
g.set_ylabels('Delay (min.)');#,color='blue',fontsize=15);


# delay-causes delays histograms limited to 250 minutes per carriers for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_15,col='Carrier',
                  hue='Cause',#hue_order=order,
                  col_wrap=4,
                  xlim=(0,250),
                  palette='muted')

g.fig.suptitle('Causes-Delays (limited to 250 min.) frequencies per Carriers\n(for flights with 15+ minutes arrival delay)')
g.map(sns.histplot,'Minutes',element ='step',fill=True,alpha=.185,bins=70)
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xlabels('Delay (min.)')
g.set_ylabels('Freq.');#,color='blue',fontsize=15);


# delay-causes delays histograms per Carriers 
# for sampled 500 data points of flights with 15+ minutes arrival delay

sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_samp_15,col='Carrier',col_order=melt_15.Carrier.unique(),
                  hue='Cause',#hue_order=order,
                  col_wrap=4,
                  #xlim=(0,250),
                  palette='muted')
title = 'Causes-Delays frequencies per Carriers\n(for sampled 500 flights with 15+ minutes arrival delay)'
g.fig.suptitle(title)
g.map(sns.histplot,'Minutes',element ='step',fill=True,alpha=.185,bins=18)
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xlabels('Delay (min.)')
g.set_ylabels('Freq.');#,color='blue',fontsize=15);


# delay-causes delays boxplots per Carriers for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_15,col='Carrier',
                  
                  col_wrap=4,
                  #xlim=(0,250),
                  palette='muted')

g.fig.suptitle('Causes-Delays boxplots per Carriers\n(for flights with 15+ minutes arrival delay)')
g.map(sns.boxplot,'Minutes','Cause',order=order,hue=melt_15['Cause'],hue_order=order,orient='h')
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes');
g.set_xlabels('Delay (min.)');#,color='blue',fontsize=15);


# delay-causes delays (limited to 150 min.) boxplots per Carriers for flights with 15+ minutes arrival delay
sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_15,col='Carrier',
                  
                  col_wrap=4,
                  ylim=(0,150),
                 )#palette='muted')

g.fig.suptitle('Causes-Delays (limited to 150 min.) boxplots per Carriers\n(for flights with 15+ minutes arrival delay)')
g.map(sns.boxplot,'Cause','Minutes',order=order,hue=melt_15['Cause'],hue_order=order,orient='v')#,palette='rocket_r')
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xticklabels(rotation = 90)
g.set_ylabels('Delay (min.)');#,color='blue',fontsize=15);


# Causes-delays violinplots per Carriers for flights with 15+ minutes arrival delay

sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_15,col='Carrier',
                  
                  col_wrap=4,
                  #ylim=(0,100),
                 )#palette='muted')

g.fig.suptitle('Causes-Delays violinplots per Carriers\n(for flights with 15+ minutes arrival delay)')
g.map(sns.violinplot,'Cause','Minutes',order=order,hue=melt_15['Cause'],hue_order=order,orient='v')#,palette='rocket_r')
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xticklabels(rotation = 90)
g.set_ylabels('Delay (min.)');#,color='blue',fontsize=15);


# Causes-delays (limited to 250 min.) violinplots per Carriers for flights with 15+ minutes arrival delay

sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_15,col='Carrier',
                  
                  col_wrap=4,
                  ylim=(-100,250),
                 )#palette='muted')

g.fig.suptitle('Causes-Delays (limited to 250 min.) violinplots per Carriers\n(for flights with 15+ minutes arrival delay)')
g.map(sns.violinplot,'Cause','Minutes',order=order,hue=melt_15['Cause'],hue_order=order,orient='v')#,palette='rocket_r')
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xticklabels(rotation = 90)
g.set_ylabels('Delay (min.)');#,color='blue',fontsize=15);


# # Causes-delays violinplots per Carriers for sampled
# # 500 flights with 15+ minutes arrival delay

# sns.set(style='whitegrid')

order = melt_15.Cause.unique()
g = sns.FacetGrid(data=melt_samp_15,col='Carrier',col_order = melt_15.Carrier.unique(),
                  
                  col_wrap=4,
                  #ylim=(-20,100),
                 )#palette='muted')

g.fig.suptitle('Causes-Delays violinplots per Carriers\n(for sampled 500 flights with 15+ minutes arrival delay)')
g.map(sns.violinplot,'Cause','Minutes',order=order,hue=melt_samp_15['Cause'],hue_order=order,)#,orient='v')#,palette='rocket_r')
g.set_titles('{col_name}')
g.add_legend(title = 'Delay causes')
g.set_xticklabels(rotation = 90)
g.set_ylabels('Delay (min.)');#,color='blue',fontsize=15);


# using catplot instead of FacetGrid directly

# sns.set(font_scale=1.19,style='whitegrid')
# g = sns.catplot(kind = 'violin',data=df_inv_15_samp,col='Carrier',col_order=df_inv_15.Carrier.unique(),col_wrap=5,
#                 y='ArrDelay',orient='v',
#                 size=3.5,
#                 gridspec_kws={'grid':1,'style':'whitegrid',})

# title = 'Arrival Delay (15+ minutes) violin plots per Carriers\nfor sampled 500 flights data'
# g.fig.suptitle(title + '\n(with vertical orientation using catplot)',y=1.05)

# g.set_titles('{col_name}');


# bivarite bar plot for delay causes time means
# for the 500 sample data points (500 rows) from df_inv_15 

plt.figure(figsize=(15,8))
sns.set(font_scale=1.19,style='ticks')

ax = sns.barplot(data = melt_samp_15, x='Cause',y='Minutes', palette='pastel')

ax.set_box_aspect(5/len(ax.patches)) # change 5 (numerator)  to modify the y/x axis ratio

sns.despine(left=True)

plt.xlabel('Cause')
plt.ylabel('Delays mean (min.)')
#ytick = [0,3.5,4,5,10,15,20]
#plt.yticks(ytick,ytick)
#plt.grid(True)

# Recalculating the causes_means
means_15_samp = melt_samp_15.groupby('Cause').mean()

# get the current tick locations and labels
locs, labels = plt.xticks() 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):

    # get the text property for the label to get the correct mean
    mean_15_samp = means_15_samp.loc[label.get_text(),'Minutes']

    # print the annotation just above the top of the bar
    plt.text(loc+.1, mean_15_samp+.2, s='{:0.3f}'.format(mean_15_samp), ha = 'left', color = 'black')

ax.set_xticklabels(causes);


# bivarite bar plot for delay causes time means
# for all data points in df_inv_15 (1,524,733  rows)  [15 minutes or more arrival delays]

plt.figure(figsize=(15,8))
sns.set(font_scale=1.19,style='ticks')

ax = sns.barplot(data = melt_15, x='Cause',y='Minutes', palette='pastel')
ax.set_box_aspect(5/len(ax.patches)) # change 5 (numerator)  to modify the y/x axis ratio

sns.despine(left=True)
plt.title('Causes-delays means \nfor the year 2008 flights',fontsize=18,
          color='#BF1B00',weight='bold',y=1.04)
plt.xlabel('Cause')
plt.ylabel('Delays mean (min.)')


ytick = [0,20,40]
plt.yticks(ytick,ytick)
plt.grid(True)

# Recalculating means
means_15 = melt_15.groupby('Cause').mean()

# get the current tick locations and labels
locs, labels = plt.xticks() 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):

    # get the text property for the label to get the correct mean
    mean_15 = means_15.loc[label.get_text(),'Minutes']

    # print the annotation just above the top of the bar
    plt.text(loc+.1, mean_15+.2, s='{:0.3f}'.format(mean_15), ha = 'left', color = 'black')

ax.set_xticklabels(causes);


# # bivariate plots of Causes-delays means by Months for year 2008 flights with arrival delay of 15+ minutes
# 1.5+ million flights

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='bar',
                data=melt_15,col='Month',
                col_wrap=4,
                x='Cause',order=order,
                y='Minutes',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,facet_kws=dict(ylim=(0,60))
               )

g.fig.suptitle('Causes-delays means by year 2008 Months\nfor flights with 15+ min. arrival delay',
               color='#BF1B00',
               fontsize=21,
               x=0.331,y=1.137)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', #labels=['Carrier', 'Weather', 'National Air System', 'Security'])
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.331,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(#causes,
                  rotation=90,fontsize=13,
                 )#,weight='bold'
g.set_ylabels('Delay mean (min.)',weight='bold',fontsize=14);

ytickslabels = (g.axes[0].get_yticks()).astype('int')
g.set_yticklabels(ytickslabels,fontsize=13);#fontsize=12,color='blue',weight='bold';


# # bivariate plots of Causes-delays means by Carriers for year 2008 flights with arrival delay of 15+ minutes
# 1.5+ million flights, 20 Carriers

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='bar',
                data=melt_15,col='Carrier',
                col_wrap=5,
                x='Cause',order=order,
                y='Minutes',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,#facet_kws=dict(ylim=(0,60))
               )

g.fig.suptitle('Causes-delays means by Carriers\nfor year 2008 flights with 15+ min. arrival delay',
               color='#BF1B00',
               fontsize=21,
               x=0.35,
               y=1.125)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', 
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.35,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(#causes,
                  rotation=90,fontsize=13.5)#,weight='bold'
g.set_ylabels('Delay mean (min.)',weight='bold',fontsize=14)

ytickslabels = (g.axes[0].get_yticks()).astype('int')
g.set_yticklabels(ytickslabels,fontsize=13.5);#fontsize=12,color='blue',weight='bold';


# bivarite bar plot for delay causes time means
# for all flights of 15 minutes or more arrival delays (1,524,733  rows) 

plt.figure(figsize=(15,8))
sns.set(font_scale=1.19,style='ticks')

ax = sns.barplot(data = melt_15, x='Cause',y='ArrDelay', palette='pastel')
ax.set_box_aspect(5/len(ax.patches)) # change 5 (numerator)  to modify the y/x axis ratio

sns.despine(left=True)
plt.title('Arrival delay means Vs. Causes \nfor the year 2008',fontsize=18,
          color='#BF1B00',weight='bold',y=1.04)
plt.xlabel('Cause')
plt.ylabel('Arrival delay mean (min.)')

#ytick = [0,3.5,4,5,10,15,20]
#plt.yticks(ytick,ytick)
#plt.grid(True)

plt.yticks([])

# Recalculating the causes_means
means_15 = melt_15.groupby('Cause').mean()

# get the current tick locations and labels
locs, labels = plt.xticks() 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):

    # get the text property for the label to get the correct mean
    mean_15 = means_15.loc[label.get_text(),'ArrDelay']

    # print the annotation just above the top of the bar
    plt.text(loc+.1, mean_15+.2, s='{:0.3f} min.'.format(mean_15), ha = 'left', color = 'black',weight='bold')

ax.set_xticklabels(causes);


# # bivariate plots of Arrival-delays means per delay causes by Months for year 2008 flights with arrival delay of 15+ minutes
# 1.5+ million flights

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='bar',
                data=melt_15,col='Month',
                col_wrap=4,
                x='Cause',order=order,
                y='ArrDelay',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,#facet_kws=dict(ylim=(0,60))
               )

g.fig.suptitle('Arrival-delays (15+ min.) means per delay causes\nby Months\nfor year 2008 flights',
               color='#BF1B00',
               fontsize=21,
               x=0.331,y=1.17)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', #labels=['Carrier', 'Weather', 'National Air System', 'Security'])
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.331,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(#causes,
                  rotation=90,fontsize=13)#,weight='bold'
g.set_ylabels('Arrival delay mean (min.)',weight='bold',fontsize=14);

ytickslabels = (g.axes[0].get_yticks()).astype('int')
g.set_yticklabels(ytickslabels,fontsize=13);#fontsize=12,color='blue',weight='bold';


# # bivariate plots of Arrival-delays means per delay causes by Carriers for year 2008 flights with arrival delay of 15+ minutes
# 1.5+ million flights, 20 Carriers

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='bar',
                data=melt_15,col='Carrier',
                col_wrap=5,
                x='Cause',order=order,
                y='ArrDelay',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,#facet_kws=dict(ylim=(0,60))
                )

g.fig.suptitle('Arrival-delays (15+ min.) means per delay causes\nby Carriers\nfor year 2008 flights',
               color='#BF1B00',
               fontsize=21,
               x=0.35,
               y=1.15)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', #labels=['Carrier', 'Weather', 'National Air System', 'Security'])
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.35,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(causes,rotation=90,fontsize=13)#,weight='bold'
g.set_ylabels('Arrival delay mean (min.)',weight='bold',fontsize=14);

ytickslabels = (g.axes[0].get_yticks()).astype('int')
g.set_yticklabels(ytickslabels,fontsize=13);#fontsize=12,color='blue',weight='bold';


# preparing 'zeros_mask' for not counting ArrDelay with zero-causes delays in df_inv_15
delays_mask = df_inv_15['CarrierDelay'] + df_inv_15['WeatherDelay'] + df_inv_15['NASDelay'] + df_inv_15['SecurityDelay']
zeros_mask =(delays_mask).isin(zero_or_null)

# preparing 'zeros_mask' for not counting ArrDelay with zero-causes delays in df_inv_15_samp

samp_delays_mask =  df_inv_15_samp['CarrierDelay'] + df_inv_15['WeatherDelay'] + \
                    df_inv_15_samp['NASDelay'] + df_inv_15_samp['SecurityDelay']
samp_zeros_mask =(samp_delays_mask).isin(zero_or_null)


# bivariate plots of ArrDelay against delay causes                        ***** using sampling data for faster scattering
# for the 500 data points sample df_inv_15_samp (500 rows)
# limiting visualizing delays to 300 minutes for causes and 400 minutes arrival delay

sns.set(font_scale=1.19,style='ticks')

g = sns.PairGrid(data = df_inv_15_samp[~samp_zeros_mask],
                 x_vars=delay_causes_cols,
                 y_vars = 'ArrDelay',
                 despine=True, height=4)
g.map(sns.regplot, scatter_kws={'alpha':1.85/10}, line_kws={"color": "orange"});
#g.set(xlim=(0,300),ylim=(0,400))
g.axes[0,0].set_ylabel('Arrival delay (min.)',color='blue')

for i in range(4):
    g.axes[0][i].set_xlabel(causes[i]+ ' (min.)',color='blue');

C:\Users\UN\AppData\Local\Temp/ipykernel_2216/2063619563.py:7: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  g = sns.PairGrid(data = df_inv_15_samp[~samp_zeros_mask],


# bivariate plots of ArrDelay against delay causes
# for the 500 data points sample df_inv_15_samp (500 rows)
# limiting visualizing delays to 250 minutes

#sns.set(font_scale=1.19,style='ticks')

g = sns.PairGrid(data = df_inv_15_samp[~zeros_mask],
                 x_vars=delay_causes_cols,
                 y_vars = 'ArrDelay',
                 despine=True, height=3.5)#height=3.3,aspect=.9)
suptitle = 'Arrival delay Vs. delay causes for the year 2008 sampled flights data points of 15 min. or more arrival delays'

plt.suptitle(suptitle,fontsize=16,color='#BF1B00',weight='bold',y=1.15)

g.map(sns.regplot, scatter_kws={'alpha':1.85/10}, line_kws={"color": "orange"});
g.set(xlim=(0,300),ylim=(0,400))
g.axes[0,0].set_ylabel('Arrival delay (min.)',color='blue')

for i in range(4):
    g.axes[0][i].set_xlabel(causes[i]+ ' delay (min.)',color='blue');

C:\Users\UN\AppData\Local\Temp/ipykernel_2216/1624003397.py:7: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  g = sns.PairGrid(data = df_inv_15_samp[~zeros_mask],


# creating a function for generating different colored paired regplots

colors = (color for color in list(sns.color_palette()))

def my_reg(x,y,color=None,*args,**kwargs):
    color= next(colors)
    sns.regplot(x,y ,color=color,*args,**kwargs)
        
# ************************************************************************ previous work
# seaborn.pairplot() changing the color of each graph

#colors = iter(['xkcd:red purple', 'xkcd:pale teal', 'xkcd:warm purple',
#       'xkcd:light forest green', 'xkcd:blue with a hint of purple',
#       'xkcd:light peach', 'xkcd:dusky purple', 'xkcd:pale mauve',
#       'xkcd:bright sky blue', 'xkcd:baby poop green', 'xkcd:brownish',
#       'xkcd:moss green', 'xkcd:deep blue', 'xkcd:melon',
#       'xkcd:faded green', 'xkcd:cyan', 'xkcd:brown green',
#       'xkcd:purple blue', 'xkcd:baby shit green', 'xkcd:greyish blue'])

#colors = iter(['xkcd:sns.color_palette()[0]', 'xkcd:sns.color_palette()[1]', 'xkcd:sns.color_palette()[2]',
#               'xkcd:sns.color_palette()[3]', 'xkcd:sns.color_palette()[4]', 'xkcd:sns.color_palette()[5]',
#               'xkcd:sns.color_palette()[6]', 'xkcd:sns.color_palette()[7]', 'xkcd:sns.color_palette()[8]',
#               'xkcd:sns.color_palette()[9]'])
#colors = iter([sns.color_palette()[0],sns.color_palette()[1],sns.color_palette()[2],
#               sns.color_palette()[3],sns.color_palette()[4],sns.color_palette()[5],
#               sns.color_palette()[6],sns.color_palette()[7],sns.color_palette()[8],
#               sns.color_palette()[9]])

#colors.reset()
#colors = np.nditer([list(sns.color_palette())])#iter(sns.color_palette())

#def my_scatter(x,y, **kwargs):
#    kwargs['color'] = next(colors)
#    plt.scatter(x,y, **kwargs)

#def my_hist(x, **kwargs):
#    kwargs['color'] = next(colors)
#   plt.hist(x, **kwargs)

#iris = sns.load_dataset("iris")
#g = sns.PairGrid(iris)
#g.map_diag(my_hist)
#g.map_offdiag(my_scatter)

# * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * trying iterators
#itr.reset()
#lst = [1,2,3,4,5]
#itr = numpy.nditer([lst])itr.next()
#itr.finished
#itr.reset()
#next(itr)
#next(lst)

#colors = iter(list(sns.color_palette()))    ---- extracts points individually not the whole palette tri-tuple

#colors = np.nditer([list(sns.color_palette())])
#colors.next()
#next(colors)


# creating a function for generating different colored paired regplots
colors = (color for color in sns.color_palette())

def my_reg(x,y,color=None,*args,**kwargs):
    color= next(colors)
    sns.regplot(x,y ,color=color,*args,**kwargs)
    
# bivariate plots of ArrDelay against delay causes for the 500 data points sample df_inv_15_samp (500 rows)
# limiting visualizing delays to 250 minutes ----------------- using PairGrid with grading delay causes colors

#sns.set(font_scale=1.19,style='ticks')
g = sns.PairGrid(data = df_inv_15_samp[~zeros_mask],
                 x_vars=delay_causes_cols,
                 y_vars = 'ArrDelay',
                 despine=True, height=4,
                 palette=sns.color_palette())

suptitle = 'Arrival delay Vs. delay causes for the year 2008 sampled flights data points of 15 min. or more arrival delays'

plt.suptitle(suptitle,fontsize=21,color='#BF1B00',weight='bold',y=1.15)

g.map(my_reg,scatter_kws={'alpha':1.85/10}, line_kws={"color": "orange"})
g.set(xlim=(0,300),ylim=(0,400))

# labeling axis
g.axes[0,0].set_ylabel('Arrival delay (min.)',color='blue')
for i in range(4):
    g.axes[0][i].set_xlabel(causes[i]+ ' delay (min.)',color=sns.color_palette('dark')[i],weight='bold');

C:\Users\UN\AppData\Local\Temp/ipykernel_2216/1506443603.py:12: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  g = sns.PairGrid(data = df_inv_15_samp[~zeros_mask],
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(


# bivariate plots of ArrDelay against delay causes for the 500 data points sample df_inv_15_samp (500 rows)
# limiting visualizing delays to 250 minutes ----------------- using PairGrid with grading delay causes colors

# creating a function for generating different colored paired regplots
colors = (color for color in sns.color_palette())

def my_reg(x,y,color=None,*args,**kwargs):
    color= next(colors)
    sns.regplot(x,y ,color=color,*args,**kwargs)
    
#sns.set(font_scale=1.19,style='ticks')
colors = (color for color in sns.color_palette())
g = sns.PairGrid(data = df_inv_15_samp[~zeros_mask],
                 x_vars=delay_causes_cols,
                 y_vars = 'ArrDelay',
                 despine=True, height=3.5,
                 palette=sns.color_palette())

suptitle = 'Arrival delay Vs. delay causes for the year 2008 sampled flights data points of 15 min. or more arrival delays'

plt.suptitle(suptitle,fontsize=21,color='#BF1B00',weight='bold',y=1.15)

g.map(my_reg,scatter_kws={'alpha':1.85/10}, line_kws={"color": "orange"})

# xticklabels = [0] + [str(n/1000) + 'k' for n in g.axes[0][0].get_xticks()[1:]]
# yticklabels = [0] + [str(n/1000) + 'k' for n in g.axes[0][0].get_yticks()[1:]]

g.set(#ylabel='Arrival delay (min.)',
      xlim=(0,300),ylim=(0,400),
      #yticks = np.arange(0,4.5,1)*100,
      #xticks=np.arange(0,3.5,1)*100,
      #yticklabels=yticklabels,
      #xticklabels=xticklabels,
     )

# labeling y axis and adjusting ytick labels
g.axes[0,0].set_ylabel('Arrival delay (min.)',color='blue',fontsize=16,weight='bold')
g.axes[0,0].set_yticks(ticks = np.arange(0,4.5,1)*100)
yticklabels = [0] + [str(n/1000) + 'k' for n in g.axes[0][0].get_yticks()[1:]]
g.axes[0,0].set_yticklabels(labels = yticklabels,fontsize=16)

# labeling x axes and adjusting xtick labels
for i in range(4):
    g.axes[0][i].set_xlabel(causes[i]+ ' delay (min.)',color=sns.color_palette('dark')[i],fontsize=16,weight='bold')
    g.axes[0,i].set_xticks(ticks = np.arange(0,3.5,1)*100)
    xticklabels = [0] + [str(n/1000) + 'k' for n in g.axes[0][i].get_xticks()[1:]]
    g.axes[0,i].set_xticklabels(xticklabels,fontsize=16);

C:\Users\UN\AppData\Local\Temp/ipykernel_2216/2559357596.py:13: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  g = sns.PairGrid(data = df_inv_15_samp[~zeros_mask],
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(


#  ------------------------------------ using FacetGrid with grading delay causes colors
g = sns.FacetGrid(data = melt_samp_15,
                  hue='Cause',
                  col = 'Cause',#palette=sns.color_palette(),
                  despine=True,col_wrap=4,
                  height=5,aspect=.75)#aspect=.77)

suptitle = 'Arrival delay Vs. delay causes \nfor the year 2008 flights sampled data points of 15 min. or more arrival delays'

plt.suptitle(suptitle,fontsize=21,color='#BF1B00',weight='bold')#,y=1.15)

g.map(sns.regplot,'Minutes','ArrDelay',scatter_kws={'alpha':1.85/10}, line_kws={"color": "orange"});
g.set(xlim=(0,300),ylim=(0,400))
g.set_titles('')
g.axes[0].set_ylabel('Arrival delay (min.)',color='blue')    #when using col_wrap your index is one dimensional
                                                            # if not then two dimensions

for i in range(4):
    g.axes[i].set_xlabel(causes[i]+ ' delay (min.)',color=sns.color_palette('dark')[i],weight='bold');


# bivariate plots of ArrDelay against delay causes
# for the 500 data points sample df_inv_15_samp (500 rows)
# with log scale

sns.set(font_scale=1.19,style='ticks')

g = sns.PairGrid(data = df_inv_15_samp[~zeros_mask],
                 x_vars=delay_causes_cols,
                 y_vars = 'ArrDelay',
                 despine=True, height=4)
title = 'Arrival delay Vs. delay causes for the year 2008 sampled flights data points of 15 min. or more arrival delays' 

plt.suptitle(title+'\n (log scaled causes delays)',fontsize=16,color='#BF1B00',weight='bold',y=1.15)
g.map(sns.regplot, scatter_kws={'alpha':1.85/10}, line_kws={"color": "orange"});
# log scale
g.set(xscale='log')#,yscale='log')

g.axes[0,0].set_ylabel('Arrival delay (min.)',color='blue')

for i in range(4):
    g.axes[0][i].set_xlabel(causes[i]+ ' delay (log min.) ',color='blue');

C:\Users\UN\AppData\Local\Temp/ipykernel_2216/779458090.py:7: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  g = sns.PairGrid(data = df_inv_15_samp[~zeros_mask],


# bivariate plots of ArrDelay against delay causes
# for the 500 data points sample df_inv_15_samp (500 rows)
# with log scale

sns.set(font_scale=1.19,style='ticks')

g = sns.PairGrid(data = df_inv_15_samp[~zeros_mask],
                 x_vars=delay_causes_cols,
                 y_vars = 'ArrDelay',
                 despine=True, height=4)
title = 'Arrival delay Vs. delay causes for the year 2008 sampled flights data points of 15 min. or more arrival delays' 

plt.suptitle(title+'\n (log scaled both axes)',fontsize=16,color='#BF1B00',weight='bold',y=1.15)
g.map(sns.regplot, scatter_kws={'alpha':1.85/10}, line_kws={"color": "orange"});
# log scale
g.set(xscale='log',yscale='log')

g.axes[0,0].set_ylabel('Arrival delay (min.)',color='blue')

for i in range(4):
    g.axes[0][i].set_xlabel(causes[i]+ ' delay (log min.) ',color='blue');

C:\Users\UN\AppData\Local\Temp/ipykernel_2216/1550254202.py:7: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  g = sns.PairGrid(data = df_inv_15_samp[~zeros_mask],


# bivariate plots of ArrDelay against delay causes
# for df_inv_15, the data points of 15 minutes ArrDelay or more (1,524,733  rows)


sns.set(font_scale=1.19,style='ticks')
g = sns.PairGrid(data = df_inv_15[~zeros_mask],
                 x_vars = delay_causes_cols,
                 y_vars = 'ArrDelay',
                 despine=True,
                 height=4)

g.map(sns.regplot, scatter_kws={'alpha':1.85/10}, line_kws={"color": "#fe8589"});
g.set(xlim=(0,2500));


#  ------------------------------------ using FacetGrid with grading delay causes colors

g = sns.FacetGrid(data = melt_15 , col='Cause',hue = 'Cause',
                  height = 5,aspect=.77) #,col_wrap=2)
g.map(sns.regplot, 'Minutes','ArrDelay',line_kws={"color": "black"})
#g.add_legend()

<seaborn.axisgrid.FacetGrid at 0x2cb4aeabf10>


# boxplot of ArrDelay against delay causes
# for the 500 data points sample df_inv_15_samp (500 rows)

plt.figure(figsize=(15,8))

plt.subplot(121)
sns.set(font_scale=1.19,style='ticks')

# main plot
ax = sns.boxplot(data= melt_samp_15,
                 y='ArrDelay',x='Cause',palette='bright')
plt.title('Arrival delay Vs. Delay causes \nfor sampled data points \nof 15 min. or more delays',
          fontsize=21,color='#BF1B00',weight='bold',y=1.04)
yticks=[0,15,35,45,60,100,150,200,250,400]
plt.yticks(yticks,yticks)

#ax.set_yticks([0,35,45,50,60,100,150,200,250,400])
ax.set_ylim([0,400])
ax.set_xticklabels(causes)
plt.ylabel('Arrival delay (min.)',fontsize=18,color='blue')
plt.xlabel('Delay cause',fontsize=18,color='blue')
plt.yticks(fontsize=13)
plt.xticks(fontsize=14)
#ax.set_yticklabels([0,30,33,35,36,40,50,60,100,150,200,250,400])

#leg = ax.legend(['Carrier','Weather','National Air System','Security'],
#                loc=1,bbox_to_anchor =(1.7,-0.1),ncol=4)
#for line in leg.get_lines():
#    line.set_linewidth(7.0)
#   lgd.set_color('black') 

sns.despine(left=True)
ax.grid(True,axis ='y')

# second plot for focusing on IQR rectangles and limiting arrival delay axis to 150 minutes 
plt.subplot(122)
sns.set(font_scale=1.19,style='white')

ax = sns.boxplot(data= melt_samp_15,
                 y='ArrDelay',x='Cause',palette='bright')

sns.despine(bottom=True)
plt.yticks(fontsize=18)
ax.set_yticks([15,23,37,52,61,100,130,150,200,250])
ax.set_ylim([10,150])

ax.grid(True,axis ='y');
ax.set_xticks([])

plt.title('Highlighting IQR rectangles \non the left boxplots',fontsize=21,color='#AD1FFF')
plt.ylabel('')
plt.subplots_adjust(wspace=.2);


# bivariate boxplots of ArrDelay against delay causes
# for the 500 data points sample df_inv_15_samp (500 rows)

plt.figure(figsize=(15,8))
plt.subplot(121)
sns.set(font_scale=1.19,style='ticks')

# main violinplot
ax = sns.violinplot(data= melt_samp_15,
                 y='ArrDelay',x='Cause',palette='bright')
sns.despine(left=True)

plt.title('Arrival delay Vs. Delay causes \nfor the year 2008 flights sampled data points \nof 15 min. or more delays',
          fontsize=21,color='#BF1B00',weight='bold',y=1.04)
plt.ylabel('Arrival delay (min.)',fontsize=18,color='blue')
plt.xlabel('Delay cause',fontsize=18,color='blue')

ax.set_xticklabels(causes)
plt.xticks(fontsize=14)

ax.set_ylim([-40,400])
yticks=[-40,-20,0,15,35,51,61,100,150,200,250,400]
plt.yticks(yticks,yticks,fontsize=13)

ax.grid(True,axis ='y')

#ax.set_yticks([0,35,45,50,60,100,150,200,250,400])
#ax.set_yticklabels([0,30,33,35,36,40,50,60,100,150,200,250,400])

#leg = ax.legend(['Carrier','Weather','National Air System','Security'],
#                loc=1,bbox_to_anchor =(1.7,-0.1),ncol=4)
#for line in leg.get_lines():
#    line.set_linewidth(7.0)
#   lgd.set_color('black') 


# second plot for focusing on IQR rectangles and limiting arrival delay axis to 150 minutes 
plt.subplot(122)
sns.set(font_scale=1.19,style='white')

ax = sns.violinplot(data= melt_samp_15,
                 y='ArrDelay',x='Cause',palette='bright')

sns.despine(bottom=True)
plt.title('Highlighting the colored distribution \non the left violinplots',fontsize=21,color='#AD1FFF')

plt.xlabel('',fontsize=18,weight='bold')
plt.ylabel('')

ax.set_xticks([])

plt.yticks(fontsize=18)
ax.set_yticks([-40,-20,0,15,23,37,51,61,100,130,150])
ax.set_ylim([-40,150])

ax.grid(True,axis ='y')
plt.subplots_adjust(wspace=.2);


# boxplot of ArrDelay against delay causes
# for all data points df_inv_15 (1,524,733 rows)

plt.figure(figsize=(15,9))

plt.subplot(121)
sns.set(font_scale=1.19,style='ticks')

# main plot
ax = sns.boxplot(data= melt_15,
                 y='ArrDelay',x='Cause',palette='bright')
plt.title('Arrival delay (15+ min.) Vs. Delay causes \nfor all data points \nof 15 min. or more delays',
          fontsize=21,color='#BF1B00',weight='bold',y=1.04)
yticks=[0,100,200,400,2500]
plt.yticks(yticks,yticks)

#ax.set_yticks([0,35,45,50,60,100,150,200,250,400])
ax.set_ylim([0,2500])
#plt.yscale('log')
ax.set_xticklabels(causes)
plt.ylabel('Arrival delay (min.)',fontsize=18,color='blue')
plt.xlabel('Delay cause',fontsize=18,color='blue')
plt.yticks(fontsize=13)
plt.xticks(fontsize=14)
#ax.set_yticklabels([0,30,33,35,36,40,50,60,100,150,200,250,400])

#leg = ax.legend(['Carrier','Weather','National Air System','Security'],
#                loc=1,bbox_to_anchor =(1.7,-0.1),ncol=4)
#for line in leg.get_lines():
#    line.set_linewidth(7.0)
#   lgd.set_color('black') 

sns.despine(left=True)
ax.grid(True,axis ='y')

# second plot for focusing on IQR rectangles and limiting arrival delay axis to 150 minutes 
plt.subplot(122)
sns.set(font_scale=1.19,style='white')

ax = sns.boxplot(data= melt_15,
                 y='ArrDelay',x='Cause',palette='bright')

sns.despine(bottom=True)

plt.title('Highlighting IQR rectangles \non the left boxplots',fontsize=21,color='#AD1FFF')
plt.xlabel('',fontsize=18)
plt.ylabel('')

plt.yticks(fontsize=18)
ax.set_yticks([15,28,35,45,62,69,110,150])
ax.set_ylim([10,150])
ax.set_xticks([])

ax.grid(True,axis ='y');

plt.subplots_adjust(wspace=.2);


# # bivariate plots of Arrival-delays (15+ min.) boxplots per delay causes by Months for year 2008 flights
# 1.5+ million flights

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='box',
                data=melt_15,col='Month',
                col_wrap=4,
                x='Cause',order=order,
                y='ArrDelay',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,facet_kws=dict(ylim=(0,3000))
               )

g.fig.suptitle('Arrival-delays (15+ min.) boxplots per delay causes\nby Months\nfor year 2008 flights',
               color='#BF1B00',
               fontsize=21,
               x=0.331,y=1.17)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', #labels=['Carrier', 'Weather', 'National Air System', 'Security'])
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.331,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(causes,rotation=90,fontsize=13)#,weight='bold'
g.set_ylabels('Arrival delay (min.)',weight='bold',fontsize=14);

ytickslabels = (g.axes[0].get_yticks()).astype('int')
ytickslabels = [0] + [str(n/1000)+'k' for n in ytickslabels[1:]]
g.set_yticklabels(ytickslabels,fontsize=13);#fontsize=12,color='blue',weight='bold';


# # bivariate plots of Arrival-delays (15+ min., limited to 150) boxplots  per delay causes by Months for year 2008 flights
# 1.5+ million flights

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='box',
                data=melt_15,col='Month',
                col_wrap=4,
                x='Cause',order=order,
                y='ArrDelay',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,facet_kws=dict(ylim=(0,150))
               )

g.fig.suptitle('Arrival-delays (15+ min., limited to 150) boxplots per delay causes\nby Months\nfor year 2008 flights',
               color='#BF1B00',
               fontsize=21,
               x=0.331,y=1.17)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', #labels=['Carrier', 'Weather', 'National Air System', 'Security'])
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.331,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(causes,rotation=90,fontsize=13)#,weight='bold'
g.set_ylabels('Arrival delay (min.)',weight='bold',fontsize=14);

ytickslabels = (g.axes[0].get_yticks()).astype('int')
g.set_yticklabels(ytickslabels,fontsize=13);#fontsize=12,color='blue',weight='bold';


# investigating ArrDelay mean per delay causes for 'Jun' and 'Jul'
melt_15[melt_15.Month.isin(['Jun','Jul'])].groupby(['Month','Cause'])['ArrDelay'].median().dropna().to_frame('ArrDelay mean')


# # bivariate plots of Arrival-delays (15+ min.) boxplots per delay causes by Carriers for year 2008 flights 
# 1.5+ million flights, 20 Carriers

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='box',
                data=melt_15,col='Carrier',
                col_wrap=5,
                x='Cause',order=order,
                y='ArrDelay',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,facet_kws=dict(ylim=(0,3000))
                )

g.fig.suptitle('Arrival-delays (15+ min.) boxplots per delay causes\nby Carriers\nfor year 2008 flights',
               color='#BF1B00',
               fontsize=21,
               x=0.35,
               y=1.15)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', #labels=['Carrier', 'Weather', 'National Air System', 'Security'])
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.35,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(causes,rotation=90,fontsize=13)#,weight='bold'
g.set_ylabels('Arrival delay (min.)',weight='bold',fontsize=14);

ytickslabels = (g.axes[0].get_yticks()).astype('int')
ytickslabels = [0] + [str(n/1000)+'k' for n in ytickslabels[1:]]
g.set_yticklabels(ytickslabels,fontsize=13);#fontsize=12,color='blue',weight='bold';


# # bivariate plots of Arrival-delays (15+ min., limited to 150) boxplots  per delay causes by Carriers for year 2008 flights
# 1.5+ million flights, 20 Carriers

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='box',
                data=melt_15,col='Carrier',
                col_wrap=5,
                x='Cause',order=order,
                y='ArrDelay',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,facet_kws=dict(ylim=(0,200))
                )

g.fig.suptitle('Arrival-delays (15+ min., limited to 150) boxplots per delay causes\nby Carriers\nfor year 2008 flights',
               color='#BF1B00',
               fontsize=21,
               x=0.35,
               y=1.15)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', #labels=['Carrier', 'Weather', 'National Air System', 'Security'])
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.35,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(causes,rotation=90,fontsize=13)#,weight='bold'
g.set_ylabels('Arrival delay (min.)',weight='bold',fontsize=14);

ytickslabels = (g.axes[0].get_yticks()).astype('int')
#ytickslabels = [0] + [str(n/1000)+'k' for n in ytickslabels[1:]]
g.set_yticklabels(ytickslabels,fontsize=13);#fontsize=12,color='blue',weight='bold';


# bivariate violinplots of ArrDelay against delay causes
# for all data points df_inv_15 (1,524,733 rows)

plt.figure(figsize=(15,8))
plt.subplot(121)
sns.set(font_scale=1.19,style='ticks')

# main violinplot
ax = sns.violinplot(data= melt_15,
                 y='ArrDelay',x='Cause',palette='bright')
yticks=[0,100,200,400,2500]
plt.yticks(yticks,yticks)
plt.title('Arrival delay Vs. Delay causes \nfor all data points \nof 15 min. or more delays',
          fontsize=21,color='#BF1B00',weight='bold',y=1.04)

#ax.set_yticks([0,35,45,50,60,100,150,200,250,400])
ax.set_ylim([0,2500])
ax.set_xticklabels(causes)
plt.ylabel('Arrival delay (min.)',fontsize=18,color='blue')
plt.xlabel('Delay cause',fontsize=18,color='blue')
plt.yticks(fontsize=13)
plt.xticks(fontsize=14)
#ax.set_yticklabels([0,30,33,35,36,40,50,60,100,150,200,250,400])

#leg = ax.legend(['Carrier','Weather','National Air System','Security'],
#                loc=1,bbox_to_anchor =(1.7,-0.1),ncol=4)
#for line in leg.get_lines():
#    line.set_linewidth(7.0)
#   lgd.set_color('black') 

sns.despine(left=True)
ax.grid(True,axis ='y')

# second plot for focusing on IQR rectangles and limiting arrival delay axis to 150 minutes 
plt.subplot(122)
sns.set(font_scale=1.19,style='white')

ax = sns.violinplot(data= melt_15,
                    y='ArrDelay',x='Cause',palette='bright')

sns.despine(bottom=True)
plt.title('Highlighting the colored distribution \non the left violinplots',fontsize=21,color='#AD1FFF')

plt.xlabel('',fontsize=18,weight='bold')
plt.ylabel('')
ax.set_ylim([-10,225])
plt.yticks(fontsize=18)
ax.set_yticks([0,15,28,45,62,69,110,150,225])
ax.set_xticks([])

ax.grid(True,axis ='y');
plt.subplots_adjust(wspace=.2);


# # bivariate plots of Arrival-delays (15+ min.) violinlots per delay causes by Months for year 2008 flights
# 1.5+ million flights

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='violin',
                data=melt_15,col='Month',
                col_wrap=4,
                x='Cause',order=order,
                y='ArrDelay',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,facet_kws=dict(ylim=(0,3000))
               )

g.fig.suptitle('Arrival-delays (15+ min.) violinplots per delay causes\nby Months\nfor year 2008 flights',
               color='#BF1B00',
               fontsize=21,
               x=0.331,y=1.17)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', #labels=['Carrier', 'Weather', 'National Air System', 'Security'])
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.331,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(causes,rotation=90,fontsize=13)#,weight='bold'
g.set_ylabels('Arrival delay (min.)',weight='bold',fontsize=14);

ytickslabels = (g.axes[0].get_yticks()).astype('int')
ytickslabels = [0] + [str(n/1000)+'k' for n in ytickslabels[1:]]
g.set_yticklabels(ytickslabels,fontsize=13);#fontsize=12,color='blue',weight='bold';


# # bivariate plots of Arrival-delays (15+ min., limited to 150) violinplots  per delay causes by Months for year 2008 flights 
# 1.5+ million flights

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='violin',
                data=melt_15,col='Month',
                col_wrap=4,
                x='Cause',order=order,
                y='ArrDelay',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,facet_kws=dict(ylim=(0,150))
               )

g.fig.suptitle('Arrival-delays (15+ min., limited to 150) violinplots per delay causes\nby Months\nfor year 2008 flights',
               color='#BF1B00',
               fontsize=21,
               x=0.331,y=1.17)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', #labels=['Carrier', 'Weather', 'National Air System', 'Security'])
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.331,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(causes,rotation=90,fontsize=13)#,weight='bold'
g.set_ylabels('Arrival delay (min.)',weight='bold',fontsize=14);

ytickslabels = (g.axes[0].get_yticks()).astype('int')
g.set_yticklabels(ytickslabels,fontsize=13);#fontsize=12,color='blue',weight='bold';


# # bivariate plots of Arrival-delays (15+ min.) violinplots per delay causes by Carriers for year 2008 flights 
# 1.5+ million flights, 20 Carriers

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='violin',
                data=melt_15,col='Carrier',
                col_wrap=5,
                x='Cause',order=order,
                y='ArrDelay',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,#facet_kws=dict(ylim=(0,3000))
                )

g.fig.suptitle('Arrival-delays (15+ min.) violinplots per delay causes\nby Carriers\nfor year 2008 flights',
               color='#BF1B00',
               fontsize=21,
               x=0.35,
               y=1.15)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', #labels=['Carrier', 'Weather', 'National Air System', 'Security'])
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.35,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(causes,rotation=90,fontsize=13)#,weight='bold'
g.set_ylabels('Arrival delay (min.)',weight='bold',fontsize=14);

#ytickslabels = (g.axes[0].get_yticks()).astype('int')
#ytickslabels = [0] + [str(n/1000)+'k' for n in ytickslabels[1:]]
ytickslabels = [str(n/1000) + 'k' for n in g.axes[0].get_yticks()]
ytickslabels = [label.replace('0.0k','0') for label in ytickslabels]
g.set_yticklabels(ytickslabels,fontsize=13);#fontsize=12,color='blue',weight='bold';


# # bivariate plots of Arrival-delays (15+ min., limited to 175) violinplots  per delay causes by Carriers for year 2008 flights
# 1.5+ million flights, 20 Carriers

sns.set(font_scale=1.19,style='whitegrid')

order = melt_15['Cause'].unique()

# g = sns.FacetGrid(data=melt_15,col='Month',
#                   col_wrap=4,
#                   hue='Cause',hue_order= order,palette='pastel'
#                  )#,xlim=(0,300))
g = sns.catplot(kind='violin',
                data=melt_15,col='Carrier',
                col_wrap=5,
                x='Cause',order=order,
                y='ArrDelay',
                hue='Cause',hue_order= order,palette='pastel',
                height=3.5,facet_kws=dict(ylim=(0,175))
                )

g.fig.suptitle('Arrival-delays (15+ min., limited to 175) violinplots per delay causes\nby Carriers\nfor year 2008 flights',
               color='#BF1B00',
               fontsize=21,
               x=0.35,
               y=1.15)

# g.map(sns.barplot,'Cause','Minutes',order=order,)

g.set_titles('{col_name}',fontsize=18,weight='bold')

g.add_legend(title = 'Delay causes', #labels=['Carrier', 'Weather', 'National Air System', 'Security'])
             loc='upper center',
             ncol=4,
             fontsize=14,
             bbox_to_anchor=(0.35,1.073),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=16)

g.set_xlabels('Cause',weight='bold',fontsize=14)#
g.set_xticklabels(causes,rotation=90,fontsize=13,)#va='center',ha='center')#,weight='bold'
g.set_ylabels('Arrival delay (min.)',weight='bold',fontsize=14);

ytickslabels = (g.axes[0].get_yticks()).astype('int')
#ytickslabels = [0] + [str(n/1000)+'k' for n in ytickslabels[1:]]
g.set_yticklabels(ytickslabels,fontsize=13);#fontsize=12,color='blue',weight='bold';


# ArrDelay means per months for all flights
month_arr_means = df_inv[['Month','ArrDelay']].groupby('Month').mean()
month_arr_means


# plotting Month Vs ArrDelay means for all flights

month_arr_means = df_inv[~(df_inv['ArrDelay'].isin(zero_or_null))][['Month','ArrDelay']] \
                                                                  .groupby('Month').mean() \
                                                                  .sort_values(by='ArrDelay',ascending=False)

plt.figure(figsize=[15,8])

sns.pointplot(data=month_arr_means,
              x=month_arr_means.index , y='ArrDelay',
              linestyles='',color=sns.color_palette()[0])

sns.despine(left=True)
sns.set_theme(font_scale=1.19,style='ticks')

plt.title('Arrival delay means per months \n for the year 2008 flights',fontsize=21,weight='bold')
plt.xlabel('Month',fontsize=18)
plt.ylabel('Arrival delay mean (min.)',fontsize=18)

plt.ylim((0,18))
plt.grid(True,axis='both')

plt.xticks(fontsize=13.5)

# get the current tick locations and labels
locs, labels = plt.xticks(fontsize=13.5,weight='bold') 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):
    # get the text property for the label to get the correct count
    mean = month_arr_means.loc[label.get_text(),'ArrDelay']
    
    string = '{:0.2f} min.'.format(mean)
    
    # print the annotation next to the bar
    plt.text(loc, mean+.33, s=string, ha = 'center', color = 'black',fontsize=13);


# ArrDelay means by causes per Months for flights with 15+ minutes arrival delay
# using factorplot(catplot now) instead of FacetGrid to overtake ordering and legend markers issues

g = sns.factorplot(data=melt_15,kind='point',
                   x='Cause',y='ArrDelay',
                   col='Month',
                   hue='Cause',palette = 'rocket_r',
                   col_wrap=4,
                   size = 4)#,xlim=(0,300))
sns.set(font_scale=1.4,style='whitegrid')

g.fig.suptitle('Arrival Delay (15+ minutes) means by delay causes per Months for year 2008 flights (using factorplot)',
               fontsize=24,x=0.331,y=1.115)
               
#g.map(sns.pointplot,'Cause','ArrDelay', order=melt_15.Cause.unique(),
#                    dodge=.7)#,linestyles='',scale=1.15,
# plt.legend(title='Delay causes',#title_fontsize =25,
#            loc='upper center',
#            ncol=4, markerscale = 2,
#            prop={'size':18},fontsize=14,
#            bbox_to_anchor=(0.331,1.085),
#            frameon=True,framealpha=.8)
# plt.legend()
g.set_titles('{col_name}',weight='bold')
g.add_legend(title='Delay causes',#title_fontsize =25,
             loc='upper center',
             ncol=4, markerscale = 1.3,
             prop={'size':18},fontsize=14,
             bbox_to_anchor=(0.331,1.085),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=22)
g.set_xticklabels(rotation=90)
g.set_xlabels('Cause',weight='bold')
g.set_ylabels('Arrival delay mean (min.)',weight='bold');#,color='blue',fontsize=15);

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\categorical.py:3714: UserWarning: The `factorplot` function has been renamed to `catplot`. The original name will be removed in a future release. Please update your code. Note that the default `kind` in `factorplot` (`'point'`) has changed `'strip'` in `catplot`.
  warnings.warn(msg)
c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\categorical.py:3720: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(msg, UserWarning)


# ArrDelay means by causes per Months for flights with 15+ minutes arrival delay
sns.set(font_scale=1.4,style='whitegrid')
g = sns.FacetGrid(data=melt_15,col='Month',col_wrap=4,
                 height=4)#,xlim=(0,300))

g.fig.suptitle('Arrival Delay (15+ minutes) means by delay causes per Months for year 2008 flights (using FacetGrid directly)',
               fontsize=24,x=0.331,y=1.13)
g.map(sns.pointplot,'Cause','ArrDelay', order=melt_15.Cause.unique(),
                            hue=melt_15['Cause'],hue_order=melt_15.Cause.unique(),
                            palette = 'rocket_r',
                            dodge=.7)#,linestyles='',scale=1.15,

g.set_titles('{col_name}',weight='bold')
g.add_legend(title='Delay causes',#title_fontsize =25,
             loc='upper center',
             ncol=4,
             prop={'size':20},
             bbox_to_anchor=(0.331,1.085),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=22)
g.set_xticklabels(rotation=90)
g.set_xlabels('Cause',weight='bold')
g.set_ylabels('Arrival delay mean (min.)',weight='bold');#,color='blue',fontsize=15);


# ArrDelay means per Airlines

airlines_arr_means = df_inv[~(df_inv['ArrDelay'].isin(zero_or_null))][['Carrier','ArrDelay']] \
                                                                                    .groupby('Carrier').mean() \
                                                                                    .sort_values(by='ArrDelay',ascending=False)


plt.figure(figsize=[10,12])

sns.barplot(data=airlines_arr_means, y=airlines_arr_means.index , x='ArrDelay')#,linestyles='')

sns.despine(left=True,bottom=True)
sns.set_theme(font_scale=1.19,style='ticks')

plt.title('Arrival delay means per airlines \n for the year 2008 flights',fontsize=21,weight='bold')
plt.ylabel('Airline',fontsize=18)
plt.xlabel('Arrival delay mean (min.)',fontsize=18)
plt.xticks([])

#plt.ylim((0,18))
plt.grid(True)

loc , labels = plt.xticks(fontsize=13.5)
loc , labels = plt.yticks(fontsize=13.5)
plt.axvline(color='black')

# Recalculating the flights_counts for carriers
carrier_flights_count = airline_cancelled['Carrier'].value_counts()

# get the current tick locations and labels
locs, labels = plt.yticks(weight='bold') 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):
    # get the text property for the label to get the correct count
    mean = airlines_arr_means.loc[label.get_text(),'ArrDelay']
    
    string = '{:0.2f} min.'.format(mean)
    
    # print the annotation next to the bar
    plt.text(mean+.3, loc, s=string, va = 'center', color = 'black',fontsize=12.5);


# ArrDelay means by causes per Carriers for flights with 15+ minutes arrival delay
sns.set(font_scale=1.4,style='whitegrid')

g = sns.FacetGrid(data=melt_15,col='Carrier',col_wrap=5,
                 height=4)#,xlim=(0,300))

g.fig.suptitle('Arrival Delay (15+ minutes) means by delay causes per Carriers for year 2008 flights',
               fontsize=24,x=0.331,y=1.13)
g.map(sns.pointplot,'Cause','ArrDelay', order=melt_15.Cause.unique(),
                            hue=melt_15['Cause'],hue_order=melt_15.Cause.unique(),
                            palette = 'rocket_r',
                            dodge=.7)#,linestyles='',scale=1.15,

g.set_titles('{col_name}',weight='bold')
g.add_legend(title='Delay causes',#title_fontsize =25,
             loc='upper center',
             ncol=4,
             prop={'size':20},
             bbox_to_anchor=(0.331,1.085),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=22)
g.set_xticklabels(rotation=90)
g.set_xlabels('Cause',weight='bold')
g.set_ylabels('Arrival delay mean (min.)',weight='bold');#,color='blue',fontsize=15);


print('\n','Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security')

airline_cancelled

 Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security


# Months Vs. flight cancellation times

plt.figure(figsize=(18,8))

ax = sns.countplot(data=airline_cancelled,
                   x='Month', order = airline_cancelled['Month'].value_counts().index,
                   #hue='CancellationCode',hue_order=['A','B','C','D'],
                   palette='pastel');
sns.despine()

plt.title('Flights cancellation times per months \nfor the year 2008 flights',
          fontsize=21,color='#BF1B00',weight='bold',y=1.04)

plt.xlabel('Month',fontsize=18,weight='bold')
plt.ylabel('Cancellation times',fontsize=18,weight='bold')

plt.xticks(fontsize=12.5,weight='bold')
plt.yticks(np.arange(0,23,2) * 1000,
           [0]+[str(n)+'k' for n in np.arange(2,23,2)],
           #[0,'2k','4k','6k','8k','10k','12k','14k','16k','18k','20k','22k'],
           fontsize=13.5,weight='bold')

#plt.grid(True,axis='y')

months_canc = airline_cancelled['Month'].value_counts()

# get the current tick locations and labels
locs, labels = plt.xticks() 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):
    # get the text property for the label to get the correct count
    count = months_canc[label.get_text()]
    
    string = f"{count:,}"
    
    # print the annotation next to the bar
    plt.text(loc, count+300, s=string, ha = 'center', color = 'black',fontsize=12.5,weight='bold');


# cancellation times per airline

plt.figure(figsize=(12,15))

ax = sns.countplot(data=airline_cancelled,
                   y='Carrier',order=airline_cancelled['Carrier'].value_counts().index,
                   #hue='CancellationCode',hue_order=['A','B','C','D'],
                   color=sns.color_palette()[7]);

sns.despine(top=False,bottom=True)


plt.title('Airlines flights cancellation-times \nfor the year 2008 flights',
          fontsize=21,color='#BF1B00',weight='bold',y=1.04)
plt.xlabel('')
plt.ylabel('Carrier',fontsize=16,weight='bold')

ax.xaxis.tick_top()

plt.xticks(np.arange(0,19000,2000),
           [0] + [str(n)+'k' for n in range(2,20,2)],
           #[0,'2k','4k','6k','8k','10k','12k','14k','16k','18k'],
           fontsize=12.5,weight='bold')

# Recalculating the flights_counts for carriers
carrier_flights_count = airline_cancelled['Carrier'].value_counts()

# get the current tick locations and labels
locs, labels = plt.yticks(weight='bold') 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):
    # get the text property for the label to get the correct count
    count = carrier_flights_count[label.get_text()]
    
    string = f"{count:,}"

    # print the annotation next to the bar
    plt.text(count+300, loc, s=string, va = 'center', color = 'black',fontsize=12.5);
    
plt.grid(True,axis='x');


# Months flights cancellation Vs. delay causes times
plt.figure(figsize=(20,8))

ax = sns.countplot(data=airline_cancelled,
                   x='Month',
                   hue='CancellationCode',hue_order=['A','B','C','D'],
                   palette='pastel');
sns.despine()

plt.title('Flights cancellation-causes times per months \nfor the year 2008 flights',
          fontsize=21,color='#BF1B00',weight='bold',y=1.04)

plt.xlabel('Months',fontsize=18,weight='bold')
plt.ylabel('Freq.',fontsize=18,weight='bold')

#plt.xticks(np.arange(0,12),months)
plt.xticks(fontsize=13.5,weight='bold')
plt.yticks(np.arange(0,12000,2000),[0,'2k','4k','6k','8k','10k'],fontsize=13.5,weight='bold')

legend = plt.legend(title='Causes',
                    title_fontsize=18,ncol=4,loc='lower center',bbox_to_anchor=(.5,-.35),
                    #edgecolor='black',prop={'size': 22},
                    labels=['Carrier','Weather','National Air System','Security'],#causes,
                    fontsize=13,prop={'size':18})#'weight':'bold',
plt.setp(legend.get_title(),fontsize=18,weight='bold')                 #adjusting legend title
#plt.grid(True,axis='y')

# drawing max and min lines
max_min_lines = {'carrier':(7312,2097),
                 'weather':(10042,600),
                 'NAS':(4470,552)}
colors = (color for color in list(sns.color_palette()))
for m in max_min_lines:
    max,min = max_min_lines[m]
    color=next(colors)
    plt.axhline(max,color=color,linewidth=1.2, linestyle ='-')
    plt.axhline(min ,color=color,linewidth=1.2, linestyle ='-');

#'-', '--', '-.', ':', 'None', ' ', '', 'solid', 'dashed', 'dashdot', 'dotted'

print('\nHint: Colored lines indicate the Max and Min frequencies');

Hint: Colored lines indicate the Max and Min frequencies


# Flights cancellation times by delay causes per Months 
sns.set(font_scale=1.4,style='whitegrid')
g = sns.FacetGrid(data=airline_cancelled,col='Month',col_wrap=4,
                 height=4)#,xlim=(0,300))

g.fig.suptitle('Flights cancellation times by delay causes per Months for year 2008 flights (using FacetGrid directly)',
               fontsize=24,x=0.331,y=1.13)
g.map(sns.countplot,'CancellationCode', order=['A','B','C','D'],
                            hue=airline_cancelled['CancellationCode'],hue_order=['A','B','C','D'],
                            palette = 'rocket_r',
                            dodge=.7)#,linestyles='',scale=1.15,

g.set_titles('{col_name}',weight='bold')
g.add_legend(labels=['Carrier', 'Weather', 'National Air System', 'Security'],
             title='Delay causes',#title_fontsize =25,
             loc='upper center',
             ncol=4,
             prop={'size':20},
             bbox_to_anchor=(0.331,1.085),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=22)
g.set_xticklabels(causes,rotation=90)
g.set_xlabels('Cause',weight='bold')
g.set_ylabels('Freq.',weight='bold')

ytickslabels = (g.axes[0].get_yticks()).astype('int')
ytickslabels = [0] + [str(int(n/1000))+'k' for n in ytickslabels[1:]]
g.set_yticklabels(ytickslabels,weight='bold');#,color='blue',fontsize=15);

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:132: UserWarning: You have mixed positional and keyword arguments, some input may be discarded.
  figlegend = self.fig.legend(handles, labels, **kwargs)


airline_cancelled_dummies = pd.get_dummies(airline_cancelled,columns=['CancellationCode'],prefix='',prefix_sep='')

print('\n','Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security')

airline_cancelled_dummies.groupby('Month').sum()

 Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security


plt.figure(figsize=(12,15))

ax = sns.countplot(data=airline_cancelled,
                   y='Carrier',order = airline_cancelled['Carrier'].value_counts().index,
                   hue='CancellationCode',hue_order=['A','B','C','D'],
                   palette='pastel');

sns.despine(top=False,bottom=True)
#plt.grid(True,axis='x')
title = 'Flights cancellation-causes frequencies by airlines \nfor the year 2008'
plt.title(title + '\n(from highest total cancelled flights per airline)',
          fontsize=21,color='#BF1B00',weight='bold',y=1.04)
plt.xlabel('')
plt.ylabel('Carrier',fontsize=16,weight='bold')

ax.xaxis.tick_top()
#ax.xaxis.set_label_position('top')  
#ax.set_xlabel('X LABEL')
#plt.xticks(rotation=80,weight='bold')
plt.xticks(np.arange(0,12,2)*1000,
           [0]+[str(n)+ 'k' for n in np.arange(2,12,2)],
           #[0,'2k','4k','6k','8k','10k'],
           fontsize=12.5,weight='bold')
plt.legend(title='Cancellation causes',title_fontsize=18,
           loc='center right',edgecolor='black', #prop={'size': 22},
           labels=['Carrier','Weather','National Air System','Security'],
           bbox_to_anchor=(1.2,.5),fontsize=16)
# Recalculating the flights_counts for carriers
#carrier_flights_count = airline_cancelled['Carrier'].value_counts()
#carrier_flights_count_whole = df_inv['Carrier'].value_counts()   # to be used in calculating percentages of carrier flights
                                                                # counts with delays of 15 minutes or more 
                                                                # against carrier total flights number 
# get the current tick locations and labels
locs, labels = plt.yticks(weight='bold') 

# loop through each pair of locations and labels
#for loc, label in zip(locs, labels):
#    # get the text property for the label to get the correct count
#    count = carrier_flights_count[label.get_text()]
#    count_whole = carrier_flights_count_whole[label.get_text()]
#    
#    percentage_value = 100*count/count_whole
#    string = str(f"{count:,}") + ' ({:0.1f}% of total)'.format(percentage_value)# flights
#
#    # print the annotation next to the bar
#    plt.text(count+3000, loc, s=string, va = 'center', color = 'black',fontsize=12.5);

# drawing max and min lines
max_min_lines = {'carrier':(9609,42),
                 'weather':(8408,0),
                 'NAS':(6275,0)}
colors = (color for color in list(sns.color_palette()))
for m in max_min_lines:
    
    max,min = max_min_lines[m]
    color=next(colors)
    plt.axvline(max,color=color,linewidth=1.7, linestyle ='-')
    plt.axvline(min ,color=color,linewidth=1.7, linestyle ='-');

#'-', '--', '-.', ':', 'None', ' ', '', 'solid', 'dashed', 'dashdot', 'dotted'

print('\nHint: Colored lines indicate the Max and Min frequencies');

Hint: Colored lines indicate the Max and Min frequencies


#  Flights cancellation times by delay causes per Carriers
sns.set(font_scale=1.4,style='whitegrid')

g = sns.FacetGrid(data=airline_cancelled,col='Carrier',col_wrap=5,
                 height=4)#,xlim=(0,300))

g.fig.suptitle('Flights cancellation times by delay causes per Carriers for year 2008 flights',
               fontsize=24,x=0.331,y=1.13)
g.map(sns.countplot,'CancellationCode', order=['A','B','C','D'],
                            hue=airline_cancelled['CancellationCode'],hue_order=['A','B','C','D'],
                            palette = 'rocket_r',
                            dodge=.7)#,linestyles='',scale=1.15,

g.set_titles('{col_name}',weight='bold')
g.add_legend(labels=['Carrier', 'Weather', 'National Air System', 'Security'],
             title='Delay causes',#title_fontsize =25,
             loc='upper center',
             ncol=4,
             prop={'size':20},
             bbox_to_anchor=(0.331,1.085),
             frameon=True,framealpha=.8)
plt.setp(g._legend.get_title(),fontsize=22)
g.set_xticklabels(causes,rotation=90)
g.set_xlabels('Cause',weight='bold')
g.set_ylabels('Freq.',weight='bold')#,color='blue',fontsize=15);

ytickslabels = (g.axes[0].get_yticks()).astype('int')
ytickslabels = [0] + [str(int(n/1000))+'k' for n in ytickslabels[1:]]
g.set_yticklabels(ytickslabels,weight='bold');

c:\users\un\appdata\local\programs\python\python39\lib\site-packages\seaborn\axisgrid.py:132: UserWarning: You have mixed positional and keyword arguments, some input may be discarded.
  figlegend = self.fig.legend(handles, labels, **kwargs)


# flights cancellation-causes times per airlines

airline_cancelled_dummies = pd.get_dummies(data=airline_cancelled,
                                           columns=['CancellationCode'],
                                           prefix='',
                                           prefix_sep='')
airline_cancelled_dummies.drop(columns='Month',inplace=True)
#airline_cancelled_dummies.rename(columns={'A':'Carrier cancellation',
#                                          'B':'Weather cancellation',
#                                          'C':'National Air system cancellation',
#                                          'D':'Security cancellation'},
#                                 inplace=True)

print('\n','Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security')
# 
airline_cancelled_dummies.groupby('Carrier').sum()                  # (A = carrier, B = weather, C = NAS, D = security)

 Cancellation Codes: A = Carrier , B = Weather , C = National Air system , D = Security


# Flights proportions by Months pie plots
plt.subplots(1,3)#,sharex=True,sharey=True)

plt.suptitle("Flights proportions by Months \nfor the year 2008",color='#BF1B00', y=1.03,fontsize=27)

plt.subplot(131)
# Total flights proportions by Months 
plt.title("All flights",color='#AD1FFF',y=.965, fontsize=24)

month_counts = df_inv['Month'].value_counts()

month_counts.plot(kind='pie',figsize=(23,8),#radius=7.5,
                    label='',#labels=n_list,
                    counterclock=False,startangle=90,
                    #explode=(0.1, 0.1, 0.1, 0.1),
                    autopct="%1.1f%%", colors=sns.color_palette(),fontsize=18)
plt.axis("equal")


plt.subplot(132)
# Total delayed flights proportions by Months 
plt.title("Delayed", color='#AD1FFF',y=.965, fontsize=24)

month_counts = df_inv_15['Month'].value_counts()

month_counts.plot(kind='pie',figsize=(23,8),#radius=7.5,
                    label='',#labels=n_list,
                    counterclock=False,startangle=90,
                    #explode=(0.1, 0.1, 0.1, 0.1),
                    autopct="%1.1f%%", colors=sns.color_palette(),fontsize=18)
plt.axis("equal")


plt.subplot(133)
# Total cancelled flights proportions by Months 

plt.title("Cancelled" ,color='#AD1FFF', y=.965,fontsize=24)

month_counts = airline_cancelled['Month'].value_counts()

month_counts.plot(kind='pie',figsize=(23,8),#radius=7.5,
                    label='',#labels=n_list,
                    counterclock=False,startangle=90,
                    #explode=(0.1, 0.1, 0.1, 0.1),
                    autopct="%1.1f%%", colors=sns.color_palette(),fontsize=18)
plt.axis("equal");


# Flights proportions by Months donut plots
plt.subplots(2,2)#,sharex=True,sharey=True)

plt.suptitle("Flights proportions \nby Months \nfor the year 2008",color='#BF1B00',fontsize=33,weight='bold', y=1)

plt.subplot(221)
# Total flights proportions by Months 
plt.title("All \n[7,009,724 flights]",color='#AD1FFF',y=1.068, fontsize=30,weight='bold')

month_counts = df_inv['Month'].value_counts()

month_counts.plot(kind='pie',figsize=(23,20),pctdistance=1.073, labeldistance=1.197,wedgeprops = {'width' : .25},#radius=7.5,
                    label='',#labels=n_list,
                    counterclock=False,startangle=90,
                    #explode=(0.1, 0.1, 0.1, 0.1),
                    autopct="%1.2f%%", colors=sns.color_palette(),fontsize=20)
plt.axis("equal")


plt.subplot(222)
# Total on-time flights proportions by Months 

plt.title("On-time \n[5,330,292 flights]" ,color='#AD1FFF', y=1.068,fontsize=28,weight='bold')

month_counts = on_time['Month'].value_counts()

month_counts.plot(kind='pie',figsize=(23,20),pctdistance=1.073, labeldistance=1.2,wedgeprops = {'width' : .25},##radius=7.5,
                    label='',#labels=n_list,
                    counterclock=False,startangle=90,
                    #explode=(0.1, 0.1, 0.1, 0.1),
                    autopct="%1.2f%%", colors=sns.color_palette(),fontsize=20)
plt.axis("equal");

plt.subplot(223)
# Total delayed flights proportions by Months 
plt.title("Delayed (15 min. arrival delay or more) \n[1,524,733 flights]", color='#AD1FFF',y=1.068, fontsize=28,weight='bold')

month_counts = df_inv_15['Month'].value_counts()

month_counts.plot(kind='pie',figsize=(23,20),pctdistance=1.073, labeldistance=1.2,wedgeprops = {'width' : .25},##radius=7.5,
                    label='',#labels=n_list,
                    counterclock=False,startangle=90,
                    #explode=(0.1, 0.1, 0.1, 0.1),
                    autopct="%1.2f%%", colors=sns.color_palette(),fontsize=20)
plt.axis("equal")

plt.subplot(224)
# Total cancelled flights proportions by Months 

plt.title("Cancelled \n[137,434 flights]" ,color='#AD1FFF', y=1.068,fontsize=28,weight='bold')

month_counts = airline_cancelled['Month'].value_counts()

month_counts.plot(kind='pie',figsize=(23,20),pctdistance=1.073, labeldistance=1.2,wedgeprops = {'width' : .25},##radius=7.5,
                    label='',#labels=n_list,
                    counterclock=False,startangle=90,
                    #explode=(0.1, 0.1, 0.1, 0.1),
                    autopct="%1.2f%%", colors=sns.color_palette(),fontsize=20)
plt.axis("equal")

plt.subplots_adjust(wspace=.15,hspace=.27);


# bivariate plots of Carrier against flights counts for the whole data points df_inv  (7,009,728 rows)

sns.set(font_scale=1.19,style='white')

plt.figure(figsize=(10,8))
default_color = sns.color_palette()[0]
sns.countplot(data = df_inv, y = 'Carrier',color = default_color)

plt.title('Total flights')
sns.despine(bottom=True)
plt.xticks([])
plt.xlabel('')

# Recalculating the flights_counts for carriers
carrier_flights_count = df_inv['Carrier'].value_counts()

# get the current tick locations and labels
locs, labels = plt.yticks() 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):

    # get the text property for the label to get the correct count
    count = carrier_flights_count[label.get_text()]
    # print the annotation just below the top of the bar
    plt.text(count+2000, loc, s=str(count) + ' flights', va = 'center', color = 'black');


# bivariate plots of Carrier against flights counts 
# for the data points with 15 minutes or more delays in df_inv_15 (1,524,733 rows) 

plt.figure(figsize=(10,8))
default_color = sns.color_palette()[0]
sns.countplot(data = df_inv_15, y = 'Carrier',color = 'skyblue')
sns.set(font_scale=1.19,style='white')

plt.title('Delayed flights')

sns.despine(bottom=True)
plt.xticks([])
plt.xlabel('')

# Recalculating the flights_counts for carriers
carrier_flights_count = df_inv_15['Carrier'].value_counts()
carrier_flights_count_whole = df_inv['Carrier'].value_counts()   # to be used in calculating percentages of carrier flights
                                                                # counts with delays of 15 minutes or more 
                                                                # against carrier total flights number 

# get the current tick locations and labels
locs, labels = plt.yticks() 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):

    # get the text property for the label to get the correct count
    count = carrier_flights_count[label.get_text()]
    count_whole = carrier_flights_count_whole[label.get_text()]
    
    string = str(count) + " flights ({:0.1f}% of carrier's)".format(100*count/count_whole)

    # print the annotation next to the bar
    plt.text(count+2000, loc, s=string, va = 'center', color = 'black');


# on-time flights by Carriers 

plt.figure(figsize=(8,12))
carrier_counts = on_time['Carrier'].value_counts()

sns.set(font_scale=1.19,style='white')


sns.countplot(data = on_time, y = 'Carrier',
              order = carrier_counts.index,
              color = sns.color_palette('vlag')[0],)#sns.color_palette('viridis')[-3],# 'skyblue'
sns.despine(bottom=True)
plt.title('On-time \n[5,330,292 flights]',fontsize=21,color='#AD1FFF',y=1.023)#,x=.25

plt.xlabel('')
plt.ylabel('')
#plt.ylabel('Month',weight='bold',fontsize=18)

plt.xticks([])
plt.yticks(color = default_color)

# Recalculating the total flights counts for months

carrier_flights_count_whole = df_inv['Carrier'].value_counts()   # to be used in calculating percentages of carrier flights
                                                                # counts with delays of 15 minutes or more 
                                                                # against carrier total flights number 
# get the current tick locations and labels
locs, labels = plt.yticks(weight='bold') 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):
    # get the text property for the label to get the correct count
    count = carrier_counts[label.get_text()]
    count_whole = carrier_flights_count_whole[label.get_text()]
    
    percentage_value = 100*count/count_whole
    percentage_total = 100*count/5330292
    
    string = str(f"{count:,}") + "   ({:0.2f}% of carrier's)\n{:0.2f}%".format(percentage_value,percentage_total)# flights

    # print the annotation next to the bar
    plt.text(count+3000, loc, s=string, va = 'center', color = 'black');


# bivariate plots of Carrier against flights counts
# for the whole data points df_inv (7,009,728 rows), against df_inv_15 (1,524,733  rows) 

default_color = sns.color_palette()[0]

fig, ax = plt.subplots(1, 2, sharex=True,figsize=(15,15))

plt.suptitle('Airlines flights counts and proportions\n for the year 2008'
          ,fontsize=22,color='#BF1B00',x=.45,y=1.002)

# first subplot
ax[0] = plt.subplot(1,2,1)

sns.set(font_scale=1.19,style='white')
sns.despine(bottom=True)

sns.countplot(data = df_inv_15, y = 'Carrier',
              order = df_inv_15['Carrier'].value_counts().index,
              color = 'skyblue',
              ax=ax[0])

plt.title('Delayed flights \n(15 min. delay or more)',fontsize=21,color='#AD1FFF',x=.25,y=1.03)

plt.xlabel('')
plt.ylabel('Carrier',weight='bold',fontsize=18)

plt.xticks([])
plt.yticks(color = 'brown')

# Recalculating the flights_counts for carriers
carrier_flights_count = df_inv_15['Carrier'].value_counts()
carrier_flights_count_whole = df_inv['Carrier'].value_counts()   # to be used in calculating percentages of carrier flights
                                                                # counts with delays of 15 minutes or more 
                                                                # against carrier total flights number 
# get the current tick locations and labels
locs, labels = plt.yticks(weight='bold') 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):
    # get the text property for the label to get the correct count
    count = carrier_flights_count[label.get_text()]
    count_whole = carrier_flights_count_whole[label.get_text()]
    
    percentage_value = 100*count/count_whole
    string = str(f"{count:,}") + " flights ({:0.1f}% of carrier's)".format(percentage_value)

    # print the annotation next to the bar
    plt.text(count+3000, loc, s=string, va = 'center', color = 'black');
    
##################################    
# second subplot
ax[1] = plt.subplot(1,2,2)
sns.countplot(data = df_inv, y = 'Carrier',
              order = df_inv['Carrier'].value_counts().index,
              color = default_color,
              ax=ax[1])
sns.set(font_scale=1.19,style='white')

sns.despine(bottom=True)

plt.title('Total flights count',fontsize=21,color='#AD1FFF',x=.45,y=1.04)
plt.xlabel('')
plt.ylabel('')

plt.xticks([])

# Recalculating the flights_counts for carriers
carrier_flights_count = df_inv['Carrier'].value_counts()

# get the current tick locations and labels
locs, labels = plt.yticks(weight='bold') 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):
    # get the text property for the label to get the correct count
    count = carrier_flights_count[label.get_text()]
    string = str(f"{count:,}") + ' flights'
    
    # print the annotation next to the bar
    plt.text(count+3000, loc, s= string, va = 'center', color = 'black');

    
#########    
#plt.subplots_adjust(wspace=0.2);

    #if percentage_value >;
#plt.yticks([])


# bivariate plots of Carrier against flights counts
# for the whole data points df_inv (7,009,728 rows), against df_inv_15 (1,524,733  rows) 

default_color = sns.color_palette()[0]

fig, ax = plt.subplots(1, 3, sharex=True,figsize=(16,20.5))

plt.suptitle('Airlines flights counts and proportions for the year 2008',
             fontsize=22,color='#BF1B00',x=.45,y=.97)
# first subplot
ax[0] = plt.subplot(1,3,1)

sns.set(font_scale=1.19,style='white')
sns.despine(bottom=True)


sns.countplot(data = airline_cancelled ,
              y = 'Carrier',
              order = airline_cancelled['Carrier'].value_counts().index,
              color = sns.color_palette()[1],
              ax=ax[0])

plt.title('Cancelled flights',fontsize=21,color='#AD1FFF',x=.205,y=1.03)

plt.xlabel('')
plt.ylabel('Carrier',weight='bold',fontsize=18)

plt.xticks([])
plt.yticks(color = sns.color_palette()[1],fontsize=12.5)

# Recalculating the flights_counts for carriers
carrier_flights_count = airline_cancelled['Carrier'].value_counts()
carrier_flights_count_whole = df_inv['Carrier'].value_counts()   # to be used in calculating percentages of carrier flights
                                                                # counts with delays of 15 minutes or more 
                                                                # against carrier total flights number 
# get the current tick locations and labels
locs, labels = plt.yticks(weight='bold') 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):
    # get the text property for the label to get the correct count
    count = carrier_flights_count[label.get_text()]
    count_whole = carrier_flights_count_whole[label.get_text()]
    
    percentage_value = 100*count/count_whole
    string = str(f"{count:,}") + " ({:0.1f}% of carrier's)".format(percentage_value)# flights

    # print the annotation next to the bar
    plt.text(count+3000, loc, s=string, va = 'center', color = 'black',fontsize=12.5);
    
##################################    
# second subplot
ax[1] = plt.subplot(1,3,2)

sns.set(font_scale=1.19,style='white')
sns.despine(bottom=True)

sns.countplot(data = df_inv_15, y = 'Carrier',
              order = df_inv_15['Carrier'].value_counts().index,
              color = 'skyblue',
              ax=ax[1])

plt.title('Delayed flights \n(15 min. delay or more)',fontsize=21,color='#AD1FFF',x=.25,y=1.03)

plt.xlabel('')
plt.ylabel('')
#plt.ylabel('Carrier',weight='bold',fontsize=18)

plt.xticks([])
plt.yticks(color = 'brown')

# Recalculating the flights_counts for carriers
carrier_flights_count = df_inv_15['Carrier'].value_counts()
carrier_flights_count_whole = df_inv['Carrier'].value_counts()   # to be used in calculating percentages of carrier flights
                                                                # counts with delays of 15 minutes or more 
                                                                # against carrier total flights number 
# get the current tick locations and labels
locs, labels = plt.yticks(weight='bold') 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):
    # get the text property for the label to get the correct count
    count = carrier_flights_count[label.get_text()]
    count_whole = carrier_flights_count_whole[label.get_text()]
    
    percentage_value = 100*count/count_whole
    string = str(f"{count:,}") + " ({:0.1f}% of carrier's)".format(percentage_value)# flights

    # print the annotation next to the bar
    plt.text(count+3000, loc, s=string, va = 'center', color = 'black');
    
##################################    
# third subplot
ax[2] = plt.subplot(1,3,3)
sns.countplot(data = df_inv, y = 'Carrier',
              order = df_inv['Carrier'].value_counts().index,
              color = default_color,
              ax=ax[2])
sns.set(font_scale=1.19,style='white')

sns.despine(bottom=True)

plt.title('Total flights count',fontsize=21,color='#AD1FFF',x=.45,y=1.04)
plt.xlabel('')
plt.ylabel('')

plt.xticks([])

# Recalculating the flights_counts for carriers
carrier_flights_count = df_inv['Carrier'].value_counts()

# get the current tick locations and labels
locs, labels = plt.yticks(weight='bold') 

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):
    # get the text property for the label to get the correct count
    count = carrier_flights_count[label.get_text()]
    string = str(f"{count:,}") #+ ' flights'
    
    # print the annotation next to the bar
    plt.text(count+3000, loc, s= string, va = 'center', color = 'black');

    
#########    
plt.subplots_adjust(wspace=0.95);

    #if percentage_value >;
#plt.yticks([])


df_inv


df_inv_15               # Month 	 Carrier 	 ArrDelay       CarrierDelay	WeatherDelay	NASDelay	SecurityDelay


# 500 sample
df_inv_15_samp.reset_index(drop=True)


melt_15                       #  Month	Carrier		ArrDelay	Cause	Minutes


melt_samp_15                       #  Month	Carrier		ArrDelay	Cause	Minutes


# correlation plot for ArrDelay and delay causes,
# for arrival delays of 15 minutes or more df_inv_15 (1,524,733 rows)

delays_cols = ['ArrDelay','CarrierDelay','WeatherDelay','NASDelay','SecurityDelay']

plt.figure(figsize = [8, 5])
sns.heatmap(df_inv_15[~zeros_mask][delays_cols].corr(),
            annot = True, fmt = '.3f',
            cmap = 'vlag_r', center = 0);


# exploring delays columns in "melt_samp_15" sampled dta of flights with 15+ min. arrival delay
    
sns.pairplot(melt_samp_15, vars=['ArrDelay','Minutes'], diag_kind="kde",
             hue="Cause",
             palette=sns.color_palette('colorblind')[:4]);


# exploring delays columns in "df_inv_15_samp" sampled dta of flights with 15+ min. arrival delay

#(using pairplot)
    
sns.pairplot(df_inv_15_samp, vars=delays_cols, diag_kind="kde");#,hue="Cause")


# PairedGrid for delays columns in "df_inv_15_samp" sampled dta of flights with 15+ min. arrival delay

# (using PairedGrid)
    
g = sns.PairGrid(df_inv_15_samp, vars=delays_cols)#,hue="Cause")
g.map_diag(sns.histplot)
g.map_offdiag(sns.scatterplot);
#g.add_legend();


# Different-causes delays Vs. Arrival delays (for sampled data of flights with 15+ min. arrival delay)

sns.scatterplot(data=melt_samp_15, x='Minutes',y='ArrDelay',hue="Cause")
plt.xlabel('Cause-delays (min.)')
plt.legend();


# Different-causes delays Vs. Arrival delays (for sampled data of flights with 15+ min. arrival delay)

sns.set(font_scale=1.19,style='whitegrid')
g = sns.PairGrid(melt_samp_15, x_vars='Minutes',y_vars='ArrDelay',hue="Cause",height=5,aspect=1.5)
g.map_diag(sns.histplot)
g.map_offdiag(sns.scatterplot)
g.add_legend()


plt.title('Different-causes delays Vs. Arrival delays\n(for sampled data of flights with 15+ min. arrival delay)\n')
plt.xlabel('Cause-delays (min.)')
plt.ylabel('Arrival delays (min.)')
#g.set_yticks(ticks = np.arange(0,3,0.5)*1000)
#yticklabels = [0] + [str(n/1000) + 'k' for n in g.axes[0][0].get_yticks()[1:]]
yticklabels = [str(n/1000) + 'k' for n in g.axes[0][0].get_yticks()]
yticklabels = [label.replace('0.0k','0') for label in yticklabels]
g.axes[0,0].set_yticklabels(labels = yticklabels)

xticklabels = [str(n/1000) + 'k' for n in g.axes[0][0].get_xticks()]
xticklabels = [label.replace('0.0k','0') for label in xticklabels]
g.axes[0,0].set_xticklabels(labels = xticklabels);#,fontsize=15.5);

C:\Users\UN\AppData\Local\Temp/ipykernel_2216/3920238421.py:17: UserWarning: FixedFormatter should only be used together with FixedLocator
  g.axes[0,0].set_yticklabels(labels = yticklabels)
C:\Users\UN\AppData\Local\Temp/ipykernel_2216/3920238421.py:21: UserWarning: FixedFormatter should only be used together with FixedLocator
  g.axes[0,0].set_xticklabels(labels = xticklabels);#,fontsize=15.5);


# Different-causes delays Vs. Arrival delays for all flights with 15+ min. arrival delay

g = sns.PairGrid(melt_15, x_vars='Minutes',y_vars='ArrDelay',hue="Cause",height=4)
g.map_diag(sns.histplot)
g.map_offdiag(sns.scatterplot)
g.add_legend()

plt.title('Different-causes delays Vs. Arrival delays\nfor all flights with 15+ min. arrival delay\n')
plt.xlabel('Cause-delays (min.)')
plt.ylabel('Arrival delays (min.)');


# Different-causes delays Vs. Arrival delays for all flights with 15+ min. arrival delay (polished)

sns.set(font_scale=1.19,style='ticks')
g = sns.PairGrid(melt_15, x_vars='Minutes',y_vars='ArrDelay',
                 hue="Cause",#palette=sns.color_palette(n_colors=4),
                 height=5,aspect=1)
g.map_diag(sns.histplot)
g.map_offdiag(sns.scatterplot)
g.add_legend()

plt.title('Different-causes delays Vs. Arrival delays\nfor all flights with 15+ min. arrival delay\n')
plt.xlabel('Cause-delays (min.)')
plt.ylabel('Arrival delays (min.)')

yticklabels = [str(n/1000) + 'k' for n in g.axes[0][0].get_yticks()]
yticklabels = [label.replace('0.0k','0') for label in yticklabels]
g.axes[0,0].set_yticklabels(labels = yticklabels)

xticklabels = [str(n/1000) + 'k' for n in g.axes[0][0].get_xticks()]
xticklabels = [label.replace('0.0k','0') for label in xticklabels]
g.axes[0,0].set_xticklabels(labels = xticklabels);

C:\Users\UN\AppData\Local\Temp/ipykernel_2216/772185180.py:17: UserWarning: FixedFormatter should only be used together with FixedLocator
  g.axes[0,0].set_yticklabels(labels = yticklabels)
C:\Users\UN\AppData\Local\Temp/ipykernel_2216/772185180.py:21: UserWarning: FixedFormatter should only be used together with FixedLocator
  g.axes[0,0].set_xticklabels(labels = xticklabels);


plt.figure(figsize = [15,8])

ax = sns.pointplot(data = melt_15, x = 'Month', y = 'ArrDelay', hue = 'Cause',
                   linestyles='',dodge= 0.2,scale=1.15
                  ,palette = 'rocket_r')

ax.set_title('Arrival delays (15 min. or more) means by delay causes delays \n per Months\n for the year 2008',
             fontsize=21,weight='bold')
ax.set_xlabel('Month',fontsize=18,weight='bold')
ax.set_ylabel('Arrival delay mean (min.)',fontsize=18,weight='bold')

plt.xticks(fontsize=15,weight='bold')
plt.yticks(fontsize=15,weight='bold')

legend = ax.legend(loc = 6, ncol = 1,  framealpha = .8, title = 'Delay causes',# labels=causes,  
                   fontsize=14, bbox_to_anchor = (1,0.5),prop={'size':18},columnspacing=.8)
plt.setp(legend.get_title(),fontsize='x-large')  

sns.despine(left=True,bottom=True)
plt.grid(True);


columns={'ArrDelay':'Arrival delay mean (min.)','Minutes':'Cause delay mean (min.)'}

#melt_15[list(set(melt_15.columns) -{'Cancelled'})].groupby(['Month','Cause']).mean().rename(columns=columns) 
melt_15.groupby(['Month','Cause']).mean().rename(columns=columns)


plt.figure(figsize = [15,8])

#plt.suptitle('Arrival delay means by delay causes \n per Airlines\n for the year 2008',fontsize=21,weight='bold')

ax = sns.pointplot(data = melt_15, x = 'Carrier', y = 'ArrDelay', hue = 'Cause',
                   linestyles='',dodge= 0.2,scale=1.15
                  ,palette = 'rocket_r')
title = 'Arrival delays (15 min. or more) means by delay causes delays \n per Airlines'

ax.set_title(title+'\n for the year 2008',fontsize=21,weight='bold',y=1.18)
ax.set_xlabel('Airline',color=sns.color_palette('viridis')[-3],
              fontsize=18,weight='bold',y=-.2)
ax.set_ylabel('Arrival delay mean (min.)', color=sns.color_palette('viridis')[-3],
              fontsize=18,weight='bold',x=-.2)

plt.xticks(fontsize=15,weight='bold',rotation=90)
plt.yticks(fontsize=15,weight='bold')

#ax.xaxis.tick_top()

legend = ax.legend(title = 'Delay causes', ncol = 4,  framealpha = .8, 
                   loc='upper center',bbox_to_anchor=(.5,1.16),
                   # labels=causes,  
                   fontsize=14,prop={'size':16},columnspacing=.8)
plt.setp(legend.get_title(),fontsize='x-large')  

sns.despine(left=True,bottom=True)
plt.grid(True);


columns={'ArrDelay':'Arrival delay mean (min.)','Minutes':'Cause delay mean (min.)'}
melt_15.groupby(['Carrier','Cause']).mean().rename(columns=columns)
#melt_15[list(set(melt_15.columns) -{'Cancelled'})].groupby(['Carrier','Cause']).mean().rename(columns=columns)


plt.figure(figsize = [15,10])

#plt.suptitle('Arrival delays means by delay causes \n per Airlines\n for the year 2008',fontsize=21,weight='bold')

ax = sns.violinplot(data = melt_15, x = 'Month', y = 'ArrDelay', hue = 'Cause',
                    #order=month_arr_means.index,
                    dodge= 0.2,#scale=1.15 ,linestyles='',
                   )#palette = 'rocket_r')


title = 'Arrival delay (15 min. or more) by delay-causes violinplots\nper Months'# (from highest total arrival delay mean)'
        

ax.set_title(title+'\nfor the year 2008',fontsize=21,weight='bold',y=1.18)

ax.set_xlabel('Month',color=sns.color_palette('viridis')[-3],
              fontsize=18,weight='bold',y=-.2)
ax.set_ylabel('Arrival delay (min.)', color=sns.color_palette('viridis')[-3],
              fontsize=18,weight='bold',x=-.2)

plt.xticks(fontsize=15,weight='bold',rotation=90)
plt.yticks(np.arange(0,3000,500),
           [0] + [str(n)+'k' for n in np.arange(0.5,3,0.5)],
           #[0] + [str(n/10)+'k' for n in range(5,30,5)],
           fontsize=15,weight='bold')

#ax.xaxis.tick_top()

legend = ax.legend(title = 'Delay causes', ncol = 4,  framealpha = .8, 
                   loc='upper center',bbox_to_anchor=(.5,1.16),
                   # labels=causes,  
                   fontsize=14,prop={'size':16},columnspacing=.8)
plt.setp(legend.get_title(),fontsize='x-large', color=sns.color_palette('viridis')[-3])  

sns.despine(left=True,bottom=True)
plt.grid(True);


plt.figure(figsize = [18,10])

#plt.suptitle('Arrival delays means by delay causes \n per Airlines\n for the year 2008',fontsize=21,weight='bold')

ax = sns.violinplot(data = melt_15, x = 'Month', y = 'ArrDelay', hue = 'Cause',
                    #order=month_arr_means.index,
                    dodge= 0.2,#scale=1.15 ,linestyles='',
                   )#palette = 'rocket_r')

title = 'Arrival delay (15 min. or more) by delay-causes violinplots\n per Months'# (from highest total arrival delay mean)'
ax.set_title(title +'\n for the year 2008' + '\n (limiting to 150 min.)',
             fontsize=21,weight='bold',y=1.18)
ax.set_xlabel('Month',color=sns.color_palette('viridis')[-3],
              fontsize=18,weight='bold',y=-.2)
ax.set_ylabel('Arrival delay (min.)', color=sns.color_palette('viridis')[-3],
              fontsize=18,weight='bold',x=-.2)
plt.ylim(-10,150)
plt.xticks(fontsize=15,weight='bold')
plt.yticks(fontsize=15,weight='bold')

#ax.xaxis.tick_top()

legend = ax.legend(title = 'Delay causes', ncol = 4,  framealpha = .8, 
                   loc='upper center',bbox_to_anchor=(.5,1.16),
                   # labels=causes,  
                   fontsize=14,prop={'size':16},columnspacing=.8)
plt.setp(legend.get_title(),fontsize='x-large', color=sns.color_palette('viridis')[-3])  

sns.despine(left=True,bottom=True)
plt.grid(True);


plt.figure(figsize = [15,10])

#plt.suptitle('Arrival delays means by delay causes \n per Airlines\n for the year 2008',fontsize=21,weight='bold')

ax = sns.violinplot(data = melt_15, x = 'Carrier', y = 'ArrDelay', hue = 'Cause',
                 #order=airlines_arr_means.index,
                 dodge= 0.2,#scale=1.15 ,linestyles='',
                )#palette = 'rocket_r')

title = 'Arrival delay (15 min. or more) by delay-causes violinplots\n per Airlines' # (from highest total arrival delay mean)'
        
ax.set_title(title+'\nfor the year 2008',fontsize=21,weight='bold',y=1.18)

ax.set_xlabel('Airline',color=sns.color_palette('viridis')[-3],
              fontsize=18,weight='bold',y=-.2)
ax.set_ylabel('Arrival delay (min.)', color=sns.color_palette('viridis')[-3],
              fontsize=18,weight='bold',x=-.2)

plt.xticks(fontsize=15,weight='bold',rotation=90)

plt.yticks(np.arange(0,3000,500),
           [0] + [str(n)+'k' for n in np.arange(0.5,3,0.5)],
           #[0] + [str(n/10)+'k' for n in range(5,30,5)],
           fontsize=15,weight='bold')

#ax.xaxis.tick_top()

legend = ax.legend(title = 'Delay causes', ncol = 4,  framealpha = .8, 
                   loc='upper center',bbox_to_anchor=(.5,1.16),
                   # labels=causes,  
                   fontsize=14,prop={'size':16},columnspacing=.8)
plt.setp(legend.get_title(),fontsize='x-large', color=sns.color_palette('viridis')[-3])  

sns.despine(left=True,bottom=True)
plt.grid(True);


plt.figure(figsize = [18,10])

#plt.suptitle('Arrival delays means by delay causes \n per Airlines\n for the year 2008',fontsize=21,weight='bold')

ax = sns.violinplot(data = melt_15, x = 'Carrier', y = 'ArrDelay', hue = 'Cause',
                    order=airlines_arr_means.index,
                    dodge= 0.2,#scale=1.15 ,linestyles='',
                   )#palette = 'rocket_r')

title = 'Arrival delay (15 min. or more) by delay-causes violinplots\n per Airlines (from highest total arrival delay mean)'
ax.set_title(title + '\n for the year 2008' + '\n(limiting to 200 min.)',
             fontsize=21,weight='bold',y=1.18)
ax.set_xlabel('Airline',color=sns.color_palette('viridis')[-3],
              fontsize=18,weight='bold',y=-.2)
ax.set_ylabel('Arrival delay (min.)', color=sns.color_palette('viridis')[-3],
              fontsize=18,weight='bold',x=-.2)
plt.ylim(-75,200)
plt.xticks(fontsize=15,weight='bold',rotation=90)
plt.yticks(fontsize=15,weight='bold')

#ax.xaxis.tick_top()

legend = ax.legend(title = 'Delay causes', ncol = 4,  framealpha = .8, 
                   loc='upper center',bbox_to_anchor=(.5,1.16),
                   # labels=causes,  
                   fontsize=14,prop={'size':16},columnspacing=.8)
plt.setp(legend.get_title(),fontsize='x-large',color=sns.color_palette('viridis')[-3])  

sns.despine(left=True,bottom=True)
plt.grid(True);


sns.color_palette()


sns.color_palette('dark')


sns.color_palette('viridis')


sns.color_palette('rocket')


sns.color_palette('vlag')


fuel_econ[['trans', 'comb']].groupby(['trans']).mean().sort_values(['comb'], ascending = False).index


# Find the order in which you want to display the Facets
# For each transmission type, find the combined fuel efficiency
group_means = fuel_econ[['trans', 'comb']].groupby(['trans']).mean()

# Select only the list of transmission type in the decreasing order of combined fuel efficiency
group_order = group_means.sort_values(['comb'], ascending = False).index

# Use the argument col_order to display the FacetGrid in the desirable group_order
g = sns.FacetGrid(data = fuel_econ, col = 'trans', col_wrap = 7, col_order = group_order)
g.map(plt.hist, 'comb')


g = sns.FacetGrid(data = fuel_econ_subset, hue = 'trans_type', 
                 hue_order = ['Automatic','Manual'],size = 4, aspect = 1.5) 
g.map(sns.regplot, 'displ', 'comb', x_jitter=0.04,fit_reg=False)
plt.xlabel('Displacement (1)')
plt.ylabel('Combined Fuel Eff. (mpg)');


g = sns.FacetGrid(data = fuel_econ, col = 'VClass',row = 'trans_type',
                 margin_titles = True)
sns.set_theme(style='whitegrid')
g.map(plt.scatter, 'displ' , 'comb')
g.set_titles(row_template='{row_name}', col_template='{col_name}')


plt.figure(figsize=(12,5))
ax = sns.barplot(data = fuel_econ, x = 'trans_type', y = 'displ', hue = 'VClass')
ax.legend(loc = 8, ncol = 3, framealpha = .8, title = 'VClass')


# evenly sampled time at 200ms intervals
t = np.arange(0., 5., 0.2)

# red dashes, blue squares and green triangles
plt.plot(t, t, 'r--', t, t**2, 'bs', t, t**3, 'g^')


ax = plt.subplot()

t = np.arange(0.0, 5.0, 0.01)
s = np.cos(2*np.pi*t)
line, = plt.plot(t, s, lw=2)

plt.annotate('local max', xy=(2, 1), xytext=(3, 1.5),
             arrowprops=dict(facecolor='black', shrink=0.05),
             )


# loading in the data, sampling to reduce points plotted
fuel_econ = pd.read_csv('fuel-econ.csv')

np.random.seed(2018)
sample = np.random.choice(fuel_econ.shape[0], 200, replace = False)
fuel_econ_subset = fuel_econ.loc[sample]

# plotting the data
plt.figure(figsize = [7,4])
plt.scatter(data = fuel_econ_subset, x = 'displ', y = 'comb', c = 'co2',
            cmap = 'viridis_r')
plt.title('Fuel Efficiency and CO2 Output by Engine Size')
plt.xlabel('Displacement (l)')
plt.ylabel('Combined Fuel Eff. (mpg)')
plt.colorbar(label = 'CO2 (g/mi)');


# YOUR CODE HERE
plt.figure(figsize=(10,6))
sb.set(font_scale=2,style='white')

plt.scatter(data = pokemon,x='defense', y='speed', c = 'special-defense', cmap ='viridis_r')
plt.xlabel('Defence')
plt.ylabel('Speed')
plt.colorbar(label='Special Defense');


g = sb.FacetGrid(data = pkmn_types_subset , hue = 'type' ,size = 5 ,aspect = 1.5)
g.map(sb.regplot, 'weight' ,'height',fit_reg=True)
#g.map(plt.scatter, 'weight' ,'height')#,fit_reg=False)

#g.set(xscale='log')
g.add_legend();


# 'solutions_multiv.py' solution code

type_cols = ['type_1','type_2']
non_type_cols = pokemon.columns.difference(type_cols)
pkmn_types = pokemon.melt(id_vars = non_type_cols, value_vars = type_cols, 
                          var_name = 'type_level', value_name = 'type').dropna()

pokemon_sub = pkmn_types.loc[pkmn_types['type'].isin(['fairy','dragon'])]

# plotting
g = sb.FacetGrid(data = pokemon_sub, hue = 'type', size = 5)
g.map(plt.scatter, 'weight','height')
g.set(xscale = 'log') # need to set scaling before customizing ticks
x_ticks = [0.1, 0.3, 1, 3, 10, 30, 100, 300, 1000]
g.set(xticks = x_ticks, xticklabels = x_ticks)
g.add_legend()


# YOUR CODE HERE
sedan_classes = ['Minicompact Cars', 'Subcompact Cars', 'Compact Cars', 'Midsize Cars', 'Large Cars']

# Returns the types for sedan_classes with the categories and orderedness
# Refer - https://pandas.pydata.org/pandas-docs/version/0.23.4/generated/pandas.api.types.CategoricalDtype.html
vclasses = pd.api.types.CategoricalDtype(ordered=True, categories=sedan_classes)

# Use pandas.astype() to convert the "VClass" column from a plain object type into an ordered categorical type 
fuel_econ['VClass'] = fuel_econ['VClass'].astype(vclasses)


# MODIFY THIS CODE
g = sb.FacetGrid(data = pokemon_sub, hue = 'type',palette=['#ee99ac','#7038f8'], size = 5)
g.map(plt.scatter, 'weight','height')
g.add_legend(title='Pokemon type',fontsize=15)
g.set(xscale = 'log')

g.set_xlabels('Weight (kg) "log scale"')
g.set_ylabels('Height (m)')
plt.title("Heights and Weights for Fairy- and Dragon-type Pokemon")


#************
# set up a dictionary to map types to colors
type_colors = {'fairy': '#ee99ac', 'dragon': '#7038f8'}

# plotting
g = sb.FacetGrid(data = pokemon_sub, hue = 'type', size = 5,
                 palette = type_colors)
g.map(plt.scatter, 'weight','height')
g.set(xscale = 'log') # need to set scaling before customizing ticks
x_ticks = [0.1, 0.3, 1, 3, 10, 30, 100, 300, 1000]
g.set(xticks = x_ticks, xticklabels = x_ticks)

# add labels and titles
g.set_xlabels('Weight (kg)')
g.set_ylabels('Height (m)')
plt.title('Heights and Weights for Fairy- and Dragon-type Pokemon')
plt.legend(['Fairy', 'Dragon'], title = 'Pokemon Type')


# convert cut, color, and clarity into ordered categorical types
ordinal_var_dict = {'cut': ['Fair','Good','Very Good','Premium','Ideal'],
                    'color': ['J', 'I', 'H', 'G', 'F', 'E', 'D'],
                    'clarity': ['I1', 'SI2', 'SI1', 'VS2', 'VS1', 'VVS2', 'VVS1', 'IF']}

for var in ordinal_var_dict:
    pd_ver = pd.__version__.split(".")
    if (int(pd_ver[0]) > 0) or (int(pd_ver[1]) >= 21): # v0.21 or later
        ordered_var = pd.api.types.CategoricalDtype(ordered = True,
                                                    categories = ordinal_var_dict[var])
        diamonds[var] = diamonds[var].astype(ordered_var)
    else: # pre-v0.21
        diamonds[var] = diamonds[var].astype('category', ordered = True,
                                             categories = ordinal_var_dict[var])


# univariate plots of cut, color, and clarity grades

plt.figure(figsize=(15,9))
sns.set(font_scale=1.14,style='white')
plt.suptitle('Frequencies of cut, color, and clarity',fontsize=18)

# Cut plot
plt.subplot(1,3,1)

cut_classes = ['Ideal', 'Premium', 'Very Good', 'Good','Fair']
cclasses = pd.api.types.CategoricalDtype(ordered=True, categories=cut_classes)
diamonds['cut'] = diamonds['cut'].astype(cclasses)

sns.countplot(diamonds['cut'])
sns.despine()
plt.title('Cut quality classes',fontsize=15)
plt.xticks(rotation=15)

# Color plot
plt.subplot(1,3,2)

color_classes = ['J', 'I', 'H', 'G', 'F', 'E', 'D']
crclasses = pd.api.types.CategoricalDtype(ordered=True, categories=color_classes)
diamonds['color'] = diamonds['color'].astype(crclasses)

#plt.bar(x=diamonds['color'].value_counts().index ,height=diamonds['color'].value_counts())
sns.countplot(diamonds['color'])
sns.despine()
plt.title('Color classes',fontsize=15)
plt.ylabel('')

# Clarity plot
plt.subplot(1,3,3)

clarity_classes = ['IF','I1','SI2','SI1','VS2','VS1','VVS2','VVS1']
clclasses = pd.api.types.CategoricalDtype(ordered=True, categories=clarity_classes)
diamonds['clarity'] = diamonds['clarity'].astype(clclasses)

sns.countplot(diamonds['clarity'])
sns.despine()
plt.title('Clarity classes',fontsize=15)
plt.ylabel('')
plt.xticks(rotation=15)
#plt.tight_layout()
;


# univariate plots of cut, color, and clarity grades

plt.figure(figsize=(15,12))
sns.set(font_scale=1.14,style='white')
plt.suptitle('Frequencies of cut, color, and clarity',fontsize=18)

# Cut plot
plt.subplot(3,1,1)

cut_classes = ['Ideal', 'Premium', 'Very Good', 'Good','Fair']
cclasses = pd.api.types.CategoricalDtype(ordered=True, categories=cut_classes)
diamonds['cut'] = diamonds['cut'].astype(cclasses)

sns.countplot(diamonds['cut'])
sns.despine()
plt.title('Cut quality classes',fontsize=15)
plt.xticks(rotation=15)
plt.grid(True)

# Color plot
plt.subplot(3,1,2)

color_classes = ['J', 'I', 'H', 'G', 'F', 'E', 'D']
crclasses = pd.api.types.CategoricalDtype(ordered=True, categories=color_classes)
diamonds['color'] = diamonds['color'].astype(crclasses)

#plt.bar(x=diamonds['color'].value_counts().index ,height=diamonds['color'].value_counts())
sns.countplot(diamonds['color'])
sns.despine()
plt.title('Color classes',fontsize=15)
plt.grid(True)
#plt.ylabel('')

# Clarity plot
plt.subplot(3,1,3)

clarity_classes = ['IF','I1','SI2','SI1','VS2','VS1','VVS2','VVS1']
clclasses = pd.api.types.CategoricalDtype(ordered=True, categories=clarity_classes)
diamonds['clarity'] = diamonds['clarity'].astype(clclasses)

sns.countplot(diamonds['clarity'])
sns.despine()
plt.title('Clarity classes',fontsize=15)
#plt.ylabel('')
plt.xticks(rotation=15)
#plt.tight_layout()
plt.grid(True)
plt.subplots_adjust(left=0.1,
                    bottom=0.1, 
                    right=0.9, 
                    top=0.9, 
                    wspace=0.4, 
                    hspace=1.4);


# multivariate plot of price by cut and color, for approx. 1 carat diamonds

plt.figure(figsize=(13,8))
sns.boxplot(data=diamonds_1, x='cut',y='price',hue='color',dodge=0.7)
plt.legend(loc='upper center',ncol=7,markerscale =.6)
plt.yscale('log')


# multivariate plot of price by cut and color, for approx. 1 carat diamonds

plt.figure(figsize=(13,8))
sns.barplot(data=diamonds_1, x='cut',y='price',hue='color',dodge=0.7,palette='rocket_r')
plt.legend(loc='upper center',ncol=7, frameon=True,framealpha = .8)


plt.figure(figsize=(15,8))
sns.pointplot(data=diamonds_1, x='cut',y='price',hue='color',dodge=0.4,linestyles='')
plt.legend(loc='upper center',ncol=7, frameon=True,framealpha = .8)


# multivariate plot of price by cut and color, for approx. 1 carat diamonds

plt.figure(figsize=(13,8))
g = sns.FacetGrid(data=diamonds_1,col='color',hue='cut',col_wrap=3,size=4)
g.map(sns.pointplot, 'cut','price',dodge=0.7,linestyles='',
      order=ordinal_var_dict['cut'])
g.set_xticklabels(rotation=15)

g.add_legend(loc=4,ncol=2)#, frameon=True,framealpha = .8)


plt.figure(figsize = [12,8])

ax = sns.pointplot(data = diamonds_1c, x = 'color', y = 'price', hue = 'cut',
                   linestyles='',dodge= 0.2,scale=1.15
                  ,palette = 'rocket_r')

# special handling of tick marks following seaborn function - you might try
# doing a log transform on the data itself instead of this workaround.

ax.set_title('Price against levels of color and cut\n for diamonds of about 1 carat weight',fontsize=21)
ax.set_xlabel('Color',fontsize=18)
ax.set_ylabel('Price (log scale)',fontsize=18)

plt.xticks(fontsize=15)

ax.set_yscale('log')

plt.yticks([3000, 5000, 8000], ['3k', '5k', '8k'],fontsize=15)
ax.set_yticklabels([], minor = True)

legend = ax.legend(loc = 6, ncol = 1,  framealpha = .8, title = 'Cut', fontsize=12, bbox_to_anchor = (1,0.5) )
plt.setp(legend.get_title(),fontsize='x-large')   # 'xx-small'
#plt.rcParams['legend.title_fontsize'] = 'xx-small'

sns.despine(left=True,bottom=True)
plt.grid(True)
plt.show();

	ArrDelay	Cancelled	CarrierDelay	WeatherDelay	NASDelay	SecurityDelay
count	6.855025e+06	7.009724e+06	1.524733e+06	1.524733e+06	1.524733e+06	1.524733e+06
mean	8.168445e+00	1.960619e-02	1.577205e+01	3.039035e+00	1.716462e+01	7.497444e-02
std	3.850194e+01	1.386427e-01	4.009913e+01	1.950289e+01	3.189497e+01	1.837941e+00
min	-5.190000e+02	0.000000e+00	0.000000e+00	0.000000e+00	0.000000e+00	0.000000e+00
25%	-1.000000e+01	0.000000e+00	0.000000e+00	0.000000e+00	0.000000e+00	0.000000e+00
50%	-2.000000e+00	0.000000e+00	0.000000e+00	0.000000e+00	6.000000e+00	0.000000e+00
75%	1.200000e+01	0.000000e+00	1.600000e+01	0.000000e+00	2.100000e+01	0.000000e+00
max	2.461000e+03	1.000000e+00	2.436000e+03	1.352000e+03	1.357000e+03	3.920000e+02

	ArrDelay	CarrierDelay	WeatherDelay	NASDelay	SecurityDelay
count	1.524733e+06	1.524733e+06	1.524733e+06	1.524733e+06	1.524733e+06
mean	5.682169e+01	1.577205e+01	3.039035e+00	1.716462e+01	7.497444e-02
std	5.716356e+01	4.009913e+01	1.950289e+01	3.189497e+01	1.837941e+00
min	1.500000e+01	0.000000e+00	0.000000e+00	0.000000e+00	0.000000e+00
25%	2.300000e+01	0.000000e+00	0.000000e+00	0.000000e+00	0.000000e+00
50%	3.700000e+01	0.000000e+00	0.000000e+00	6.000000e+00	0.000000e+00
75%	6.900000e+01	1.600000e+01	0.000000e+00	2.100000e+01	0.000000e+00
max	2.461000e+03	2.436000e+03	1.352000e+03	1.357000e+03	3.920000e+02

	A	B	C	D
Month
Jan	6635.0	6711.0	3962.0	0.0
Feb	6090.0	10034.0	4470.0	2.0
Mar	6038.0	7520.0	2621.0	4.0
Apr	7312.0	1479.0	1564.0	0.0
May	3157.0	1461.0	1608.0	3.0
Jun	4251.0	3448.0	3232.0	0.0
Jul	5292.0	2612.0	2694.0	0.0
Aug	3852.0	3582.0	2401.0	0.0
Sep	2246.0	6478.0	1188.0	1.0
Oct	2097.0	600.0	552.0	0.0
Nov	2510.0	937.0	1009.0	2.0
Dec	4850.0	10042.0	2887.0	0.0

	A	B	C	D
Carrier
AirTran Airways Corporation	815.0	976.0	445.0	0.0
Alaska Airlines Inc.	1222.0	865.0	52.0	0.0
Aloha Airlines Inc.	42.0	0.0	0.0	0.0
American Airlines Inc.	9609.0	5495.0	2334.0	2.0
American Eagle Airlines Inc.	3644.0	8408.0	6275.0	4.0
Atlantic Southeast Airlines	1384.0	2283.0	1357.0	2.0
Comair Inc.	1328.0	5016.0	118.0	0.0
Continental Air Lines Inc.	746.0	2878.0	78.0	0.0
Delta Air Lines Inc.	3113.0	2283.0	1417.0	0.0
Expressjet Airlines Inc.	697.0	5424.0	3871.0	0.0
Frontier Airlines Inc.	109.0	192.0	2.0	0.0
Hawaiian Airlines Inc.	563.0	7.0	0.0	0.0
JetBlue Airways	665.0	2445.0	95.0	0.0
Mesa Airlines Inc.	5020.0	1766.0	2433.0	0.0
Northwest Airlines Inc.	1119.0	901.0	886.0	0.0
Pinnacle Airlines Inc.	4329.0	1932.0	838.0	1.0
Skywest Airlines Inc.	3162.0	4748.0	4525.0	1.0
Southwest Airlines Co.	6554.0	5167.0	666.0	2.0
US Airways Inc.	3857.0	1131.0	1594.0	0.0
United Air Lines Inc.	6352.0	2987.0	1202.0	0.0

		Arrival delay mean (min.)	Cause delay mean (min.)
Month	Cause
Jan	Carrier	56.171442	36.079481
	National Air System	54.892898	28.103122
	Security	37.524057	16.790637
	Weather	80.988028	43.981173
Feb	Carrier	58.401192	36.286006
	National Air System	57.075527	27.780957
	Security	42.727987	17.863208
	Weather	82.425008	44.660814
Mar	Carrier	56.856152	34.910855
	National Air System	55.464507	26.868463
	Security	37.348328	17.435986
	Weather	84.570406	46.547068
Apr	Carrier	55.486343	35.108070
	National Air System	50.932327	25.086923
	Security	35.063025	16.060924
	Weather	83.361501	48.953925
May	Carrier	53.620574	33.535753
	National Air System	49.617749	26.008819
	Security	38.816701	19.859470
	Weather	76.167979	41.997025
Jun	Carrier	59.820534	35.793324
	National Air System	59.588023	30.752175
	Security	40.255499	17.959391
	Weather	86.900625	47.672530
Jul	Carrier	61.506064	39.506669
	National Air System	62.192084	33.147465
	Security	43.842610	25.193858
	Weather	89.995229	55.242421
Aug	Carrier	59.367418	38.772434
	National Air System	57.919304	30.155358
	Security	37.902116	19.613757
	Weather	82.097683	46.223175
Sep	Carrier	53.721025	39.542116
	National Air System	48.372589	27.562945
	Security	37.636816	23.995025
	Weather	71.277867	50.443945
Oct	Carrier	51.009869	34.557236
	National Air System	44.460146	24.703283
	Security	37.105634	18.609155
	Weather	76.976979	42.444265
Nov	Carrier	55.038367	33.525817
	National Air System	51.180908	28.693960
	Security	37.261411	16.958506
	Weather	76.694975	38.078099
Dec	Carrier	63.262444	33.489687
	National Air System	62.661970	27.564933
	Security	41.957162	16.306560
	Weather	90.865911	44.833603

Data Expo 2008 «Airline on-time performance»

by Mostafa Abobakr

Preliminary Wrangling¶

What is the structure of your dataset?¶

What is/are the main feature(s) of interest in your dataset?¶

What features in the dataset do you think will help support your investigation into your feature(s) of interest?¶

Univariate Exploration¶

Discuss the distribution(s) of your variable(s) of interest. Were there any unusual points? Did you need to perform any transformations?¶

Of the features you investigated, were there any unusual distributions? Did you perform any operations on the data to tidy, adjust, or change the form of the data? If so, why did you do this?¶

Bivariate Exploration¶

Talk about some of the relationships you observed in this part of the investigation. How did the feature(s) of interest vary with other features in the dataset?¶

Did you observe any interesting relationships between the other features (not the main feature(s) of interest)?¶

Multivariate Exploration¶

Talk about some of the relationships you observed in this part of the investigation. Were there features that strengthened each other in terms of looking at your feature(s) of interest?¶

Were there any interesting or surprising interactions between features?¶

	Year	Month	DayofMonth	DayOfWeek	DepTime	CRSDepTime	ArrTime	CRSArrTime	UniqueCarrier	FlightNum	...	TaxiIn	TaxiOut	CancellationCode	CarrierDelay	WeatherDelay	NASDelay	SecurityDelay	LateAircraftDelay
0	2008	1	3	4	2003.0	1955	2211.0	2225	WN	335	...	4.0	8.0	NaN	NaN	NaN	NaN	NaN	NaN
1	2008	1	3	4	754.0	735	1002.0	1000	WN	3231	...	5.0	10.0	NaN	NaN	NaN	NaN	NaN	NaN
2	2008	1	3	4	628.0	620	804.0	750	WN	448	...	3.0	17.0	NaN	NaN	NaN	NaN	NaN	NaN
3	2008	1	3	4	926.0	930	1054.0	1100	WN	1746	...	3.0	7.0	NaN	NaN	NaN	NaN	NaN	NaN
4	2008	1	3	4	1829.0	1755	1959.0	1925	WN	3920	...	3.0	10.0	NaN	2.0	0.0	0.0	0.0	32.0

	Year	Month	DayofMonth	DayOfWeek	DepTime	CRSDepTime	ArrTime	CRSArrTime	UniqueCarrier	FlightNum	...	TaxiIn	TaxiOut	CancellationCode	CarrierDelay	WeatherDelay	NASDelay	SecurityDelay	LateAircraftDelay
938221	2008	2	28	4	829.0	835	1341.0	1350	F9	728	...	6.0	14.0	NaN	NaN	NaN	NaN	NaN	NaN
938223	2008	2	28	4	1427.0	1435	1640.0	1636	F9	727	...	9.0	16.0	NaN	NaN	NaN	NaN	NaN	NaN
938225	2008	2	28	4	1854.0	1807	1946.0	1902	F9	773	...	8.0	13.0	NaN	44.0	0.0	0.0	0.0	0.0
938227	2008	2	28	4	2027.0	1942	2314.0	2229	F9	780	...	10.0	13.0	NaN	1.0	0.0	44.0	0.0	0.0

	FlightNum	TailNum	Month	DayofMonth	DayOfWeek	Carrier	ArrDelay	CancellationCode	CarrierDelay	WeatherDelay	NASDelay	SecurityDelay	Origin	Dest	Distance	TaxiIn	TaxiOut
0	335	N712SW	1	3	4	Southwest Airlines Co.	-14.0	NaN	NaN	NaN	NaN	NaN	IAD	TPA	810	4.0	8.0
1	3231	N772SW	1	3	4	Southwest Airlines Co.	2.0	NaN	NaN	NaN	NaN	NaN	IAD	TPA	810	5.0	10.0
2	448	N428WN	1	3	4	Southwest Airlines Co.	14.0	NaN	NaN	NaN	NaN	NaN	IND	BWI	515	3.0	17.0
3	1746	N612SW	1	3	4	Southwest Airlines Co.	-6.0	NaN	NaN	NaN	NaN	NaN	IND	BWI	515	3.0	7.0
4	3920	N464WN	1	3	4	Southwest Airlines Co.	34.0	NaN	2.0	0.0	0.0	0.0	IND	BWI	515	3.0	10.0

	Month	Carrier	CancellationCode
count	137434	137434	137434
unique	12	20	4
top	Feb	American Eagle Airlines Inc.	B
freq	20596	18331	54904

	ArrDelay	Cancelled	CarrierDelay	WeatherDelay	NASDelay	SecurityDelay
count	5.330292e+06	5330292.0	0.0	0.0	0.0	0.0
mean	-5.748843e+00	0.0	NaN	NaN	NaN	NaN
std	1.004538e+01	0.0	NaN	NaN	NaN	NaN
min	-5.190000e+02	0.0	NaN	NaN	NaN	NaN
25%	-1.200000e+01	0.0	NaN	NaN	NaN	NaN
50%	-6.000000e+00	0.0	NaN	NaN	NaN	NaN
75%	1.000000e+00	0.0	NaN	NaN	NaN	NaN
max	1.400000e+01	0.0	NaN	NaN	NaN	NaN

	Month	Carrier	ArrDelay	Cause	Minutes
0	Jan	American Airlines Inc.	50.0	Carrier	17.0
1	Sep	Expressjet Airlines Inc.	18.0	Carrier	9.0
2	Jan	United Air Lines Inc.	24.0	Carrier	7.0
3	Oct	Skywest Airlines Inc.	35.0	Carrier	35.0
4	Oct	Atlantic Southeast Airlines	26.0	Carrier	3.0
...	...	...	...	...	...
561	Mar	JetBlue Airways	145.0	National Air System	145.0
562	Jan	Skywest Airlines Inc.	15.0	Security	15.0
563	Jan	Southwest Airlines Co.	79.0	Security	13.0
564	Dec	US Airways Inc.	24.0	Security	13.0
565	Feb	Southwest Airlines Co.	158.0	Security	8.0

	ArrDelay
Month
Jan	10.188856
Feb	13.077779
Mar	11.192365
Apr	6.807297
May	5.978448
Jun	13.266756
Jul	9.975050
Aug	6.910915
Sep	0.697733
Oct	0.415495
Nov	2.015858
Dec	16.680505